Báo cáo khoa học: 6-Phosphofructo-2-kinase and fructose-2,6-bisphosphatase in Trypanosomatidae Molecular characterization, database searches, modelling studies and evolutionary analysis pptx

brucei proteins showed most conservation in the PFK-2 domain, although one of them was predicted to be inactive due to substitution of residues respon-sible for ligating the catalyticall

in Trypanosomatidae

Molecular characterization, database searches, modelling studies and evolutionary analysis

Nathalie Chevalier1,*, Luc Bertrand2,3,*, Mark H Rider2, Fred R Opperdoes1, Daniel J Rigden4 and Paul A M Michels1

1 Research Unit for Tropical Diseases, Christian de Duve Institute of Cellular Pathology and Laboratory of Biochemistry, Universite´ catholique

de Louvain, Brussels, Belgium

2 Hormone and Metabolic Research Unit, Christian de Duve Institute of Cellular Pathology and Laboratory of Biochemistry, Universite´ catholique de Louvain, Brussels, Belgium

3 Division of Cardiology, Universite´ catholique de Louvain, Brussels, Belgium

4 School of Biological Sciences, University of Liverpool, UK

Keywords

ankyrin-repeat motif; fructose

2,6-bisphosphate; glycolysis regulation;

6-phosphofructo-2-kinase ⁄

fructose-2,6-bisphosphatase; trypanosome

Correspondence

P A M Michels, ICP-TROP 74.39, Avenue

Hippocrate 74, B-1200 Brussels, Belgium

Fax: +32 27 62 68 53

Tel: +32 27 64 74 73

E-mail: michels@bchm.ucl.ac.be

*These authors contributed equally to this

work

Database

Nucleotide sequence data are available in

the DDJB ⁄ EMBL ⁄ GenBank databases under

accession numbers AY571277 (Tb4),

AY571278 (Tb1) and AY999068 (Tb2)

(Received 7 April 2005, revised 10 May

2005, accepted 16 May 2005)

doi:10.1111/j.1742-4658.2005.04774.x

Fructose 2,6-bisphosphate is a potent allosteric activator of trypanosomatid pyruvate kinase and thus represents an important regulator of energy meta-bolism in these protozoan parasites A 6-phosphofructo-2-kinase, respon-sible for the synthesis of this regulator, was highly purified from the bloodstream form of Trypanosoma brucei and kinetically characterized By searching trypanosomatid genome databases, four genes encoding proteins homologous to the mammalian bifunctional enzyme 6-phosphofructo-2-kinase⁄ fructose-2,6-bisphosphatase (PFK-2 ⁄ FBPase-2) were found for both

T brucei and the related parasite Leishmania major and four pairs in Try-panosoma cruzi These genes were predicted to each encode a protein in which, at most, only a single domain would be active Two of the T brucei proteins showed most conservation in the PFK-2 domain, although one of them was predicted to be inactive due to substitution of residues respon-sible for ligating the catalytically essential divalent metal cation; the two other proteins were most conserved in the FBPase-2 domain The two PFK-2-like proteins were expressed in Escherichia coli Indeed, the first dis-played PFK-2 activity with similar kinetic properties to that of the enzyme purified from T brucei, whereas no activity was found for the second Interestingly, several of the predicted trypanosomatid PFK-2⁄ FBPase-2 proteins have long N-terminal extensions The N-terminal domains of the two polypeptides with most similarity to mammalian PFK-2s contain a ser-ies of tandem repeat ankyrin motifs In other proteins such motifs are known to mediate protein–protein interactions Phylogenetic analysis sug-gests that the four different PFK-2⁄ FBPase-2 isoenzymes found in Trypanosoma and Leishmania evolved from a single ancestral bifunctional enzyme within the trypanosomatid lineage A possible explanation for the evolution of multiple monofunctional enzymes and for the presence of the ankyrin-motif repeats in the PFK-2 isoenzymes is presented

Abbreviations

CDD, conserved domain databases; Fru2,6-P 2 , fructose 2,6-bisphosphate; FBPase-2, fructose-2,6-bisphosphatase; PEP, phosphoenol-pyruvate; PFK-1, 6-phosphofructo-1-kinase; PFK-2, 6-phosphofructo-2-kinase; PKA, protein kinase A; PKC, protein kinase C; TbFBPase-2, Trypanosoma brucei fructose-2,6-bisphosphatase; TbPFK-2, Trypanosoma brucei 6-phosphofructo-2-kinase.

Fructose 2,6-bisphosphate (Fru2,6-P2) is a key

regula-tor of glycolysis in almost all eukaryotes, but it is

absent from prokaryotes In animals, plants and fungi,

this sugar phosphate stimulates glycolysis via allosteric

stimulation of 6-phosphofructo-1-kinase (PFK-1) and

inhibits gluconeogenesis by acting as a negative

effec-tor of fructose-1,6-bisphosphatase [1,2] (Fig 1A) In

contrast, in protozoan organisms belonging to the

Kinetoplastida (comprising pathogenic organisms such

as Trypanosoma and Leishmania) Fru2,6-P2 is not a

stimulator of PFK-1, but rather acts on pyruvate

kinase [3–9] This latter enzyme is stimulated at

sub-micromolar concentrations, 2000-fold lower than by

fructose 1,6-bisphosphate, the usual regulator of

pyru-vate kinase activity in other organisms Similarly,

try-panosomatid fructose-1,6-bisphosphatase is insensitive

to Fru2,6-P2 This different enzyme specificity is most

likely related to the unique metabolic regulation in

Kinetoplastida (Fig 1B) The majority of glycolytic

enzymes responsible for the conversion of glucose into

3-phosphoglycerate are compartmentalized in

peroxi-some-like organelles called glycosomes [10–12] Only

the last three enzymes, phosphoglycerate mutase,

eno-lase and pyruvate kinase are present in the cytosol

The gluconeogenic enzyme fructose-1,6-bisphosphatase

also has a glycosomal localization [12] Strikingly,

gly-cosomal enzymes such as hexokinase and PFK-1 lack

the regulatory mechanisms found in most other

organ-isms, namely product inhibition and control by

meta-bolites further downstream in the pathway or by

effectors [13] In Kinetoplastida, such mechanisms

seem to be redundant as a result of the sequestering of

the enzymes within a separate compartment bounded

by a membrane with low permeability to many

meta-bolites [12,14,15] Interestingly, this compartmentation

seems to have resulted in a kind of ‘re-routing’ of reg-ulatory mechanisms (Fig 1) and cytosolic pyruvate kinase appears the most important regulated enzyme [5,6,11,12,14]

In mammalian tissues, the bifunctional enzyme 6-phosphofructo-2-kinase (PFK-2; EC 2.7.1.105)⁄ fruc-tose-2,6-bisphosphatase (FBPase-2; EC 3.1.3.46) cata-lyses both the synthesis and degradation of Fru2,6-P2 [16,17] PFK-2 and FBPase-2 activities have also been detected in Kinetoplastida In line with the activity regulation of cytosolic pyruvate kinase by Fru2,6-P2, the PFK-2 and FBPase-2 activities are localized in the cytosol of the kinetoplastid Trypanosoma brucei [4] However, these activities could be separated by partial protein purification, indicating that they reside in dis-tinct enzymes [4]

In higher animals, different tissue-specific PFK-2⁄ FBPase-2 bifunctional isoenzymes exist with kinetic properties and regulatory mechanisms related to meta-bolism (glycolysis vs gluconeogenesis) [1,2,16,17] The isoenzymes are homodimers, typically with subunit masses of 50–60 kDa They have a common structure, with the PFK-2 domain comprising most of the N-ter-minal half of the enzyme subunit and the FBPase-2 domain in the C-terminal half This central core con-tains the two catalytic activities and is well conserved Both extremities often contain regulatory phosphoryla-tion sites, involved in the fine tuning of enzyme activ-ity In plants (Arabidopsis, spinach, potato), the bifunctional enzyme sequence possesses a large N-ter-minal extension that provides a regulatory domain [18] In Saccharomyces cerevisiae, three isoforms of PFK-2⁄ FBPase-2 are present, but each displays only a single activity [19–21] Nevertheless, they are clearly homologous to the mammalian bifunctional enzyme

Fig 1 Diagrammatic representation of the regulation of carbohydrate metabolism by fructose 2,6-bisphosphate in mammalian cells (A) and Trypanosomatidae (B) Glycolysis in mammalian cells occurs in the cytosol In Trypanosomatidae, the glycolytic enzymes responsible for the conversion of glucose into 3-phosphoglycerate, and the gluconeogenic enzyme fructose-1,6-bisphosphatase are present in glycosomes Abbreviations: FBPase-1, fructose-1,6-bisphosphatase; PYK, pyruvate kinase.

In two isoforms (called PFK26 and PFK27), FBPase-2

activity appears to have been lost during evolution as

the result of a crucial substitution or major deletions,

respectively In the third form (FBP26), the PFK-2

domain has undergone multiple substitutions rendering

it inactive

For mammalian PFK-2⁄ FBPase-2 isoenzymes, it has

been shown that the functional, active form is a dimer

Although the FBPase-2 domain does not seem to be

involved in dimerization, its presence, whether active

or not, seems to be essential because a mammalian

PFK-2 domain expressed alone in bacteria forms

inac-tive aggregates [22] By contrast, the bacterially

expressed mammalian FBPase-2 domain is active as a

monomer [23] The crystal structures of the rat testis

[24] and rat liver bifunctional enzymes [25] have been

solved The subunits of the dimer are arranged in a

head-to-head fashion with the PFK-2 domains of the

two subunits being intimately associated in both testis

and liver structures The FBPase-2 domains are

inde-pendent of each other in the testis enzyme but form

contacts in the liver isozyme The fact that the PFK-2

domain is structurally related to the adenylate kinase

family, whereas the FBPase-2 domain is similar to

the phosphoglycerate mutase family [24,26], seems to

reveal that the bifunctional enzyme resulted from the

fusion of two ancestral genes

We studied some properties of PFK-2 (TbPFK-2)

purified from T brucei as well as bacterially expressed

forms of TbPFK-2 and investigated whether the

T brucei PFK-2 and FBPase-2 enzymes are

homolog-ous to their counterparts in higher eukaryotes

Results and Discussion

Purification and characterization

of T brucei PFK-2

PFK-2 was purified from pooled cytosol fractions from

the bloodstream form of T brucei stock 427 using

ion-exchange chromatography and specific elution from

Blue Sepharose with buffer containing PFK-2

sub-strates The purification was 9000-fold compared with

the activity in the initial extract and the purified

enzyme had a specific activity of 11 mUnitsÆmg)1 of

protein, which is comparable with the specific activity

of PFK-2 in preparations purified from mammalian

tissues [27] Nevertheless, several bands were visible in

Coomassie Brilliant Blue-stained gels On gel filtration,

PFK-2 activity eluted from a Superose 12 column as

a single symmetrical peak with a Mr of 76 400 (not

shown) As yeast PFK-2 has been shown to be

phos-phorylated by protein kinase A (PKA) [28,29] and

protein kinase C (PKC) [30] with accompanying chan-ges in PFK-2 activity, we tested the effect of phos-phorylation by these protein kinases on T brucei PFK-2 activity (Table 1) Treatment with PKA led to PFK-2 inactivation via a sevenfold increase in Km for fructose 6-phosphate with no change in Vmax, whereas treatment with PKC was without effect This contrasts with yeast PFK-2, in which PKA treatment led to PFK-2 activation by increasing the Vmax and lowering the Km for fructose 6-phosphate [28,29] and PKC treatment led to PFK-2 inactivation [30] Purified

T brucei PFK-2 had a pH optimum around 6 (not shown), was inhibited by phosphoenolpyruvate (PEP;

K0.5¼ 0.7 mm) and citrate (60% inhibition at 1 mm) but like heart PFK-2 [31], was rather insensitive to inhibition by glycerol 3-phosphate (20% inhibition at

2 mm)

Database searches and sequence analysis tblastn searches were performed in the databases of the three trypanosomatid genome projects (T brucei,

T cruzi and Leishmania major) using a query of mammalian and yeast bifunctional PFK-2⁄ FBPase-2 sequences Analysis of the various T brucei and

L majordatabases surprisingly revealed four homolog-ous sequences The T brucei sequences and their close homologues in L major were named Tb1⁄ Lm1, Tb2⁄ Lm2, Tb3 ⁄ Lm3 and Tb4 ⁄ Lm4 and have the data-base codes shown in Table 2 A diagrammatic repre-sentation of the four T brucei sequences is presented

in Fig 2 For each of these four isoforms, two corres-ponding sequences could be found in the T cruzi genome database (Table 2), presumably reflecting the known hybrid genotype of the strain chosen for gen-ome sequencing, as a result of genetic exchange

Table 1 Kinetic properties of PFK-2 purified from bloodstream-form

T brucei and effect of phosphorylation by protein kinases Trypano-some PFK-2 (30 lgÆmL)1) was incubated for 15 min at 30
C with

or without protein kinases (0.6 unitÆmL)1), purified and assayed as described previously [87] Aliquots were taken for PFK-2 activity measurements For the fructose 6-phosphate (F6P) saturation curves, concentrations were varied up to 30 m M For inhibition by PEP, the concentrations of fructose 6-phosphate and MgATP were

5 m M The results are the means ± SEM of three separate determi-nations, otherwise individual values are given ND, not determined.

Enzyme

KmF6P (m M )

KmATP (m M )

V max

[mUnits (mg of protein))1] Untreated 5.8 ± 1.2 0.88 7.1 ± 3.2 PKA-treated 39 ND 7.1 PKC-treated 5.8 ND 6.7

between two distantly related lineages [32] Each pair,

for example Tc1–1 and Tc1–2 corresponding to Tb1,

share 95–98% sequence identity overall Between the

three trypanosomatids, isoenzyme 1 (Tb1, Lm1, Tc1–1

and Tc1–2) representatives share 35–43% sequence

identity The corresponding figures for isoenzymes 2, 3

and 4 are 43–49, 37–52 and 33–40% (Table 2)

Mammalian and yeast enzymes usually have mole-cular masses of 50–60 kDa By comparison, most of the trypanosomatidal homologues are atypically large (Table 2) Most notable are the isoenzymes 1, which range in size from 1021 residues (112 kDa) for Tc1–1

to 2422 residues (251 kDa) for Lm1 The L major rep-resentative of the isoenzymes 4 is also particularly large (1241 residues; 132 kDa) compared with the cor-responding trypanosomal sequences at around 700–750 residues

The bulk of known homologues of the bifunctional enzymes possess both kinase and phosphatase activit-ies Nevertheless, there is a precedent for homologues having retained only a single activity in the three yeast members of the family In order to predict likely activ-ities for the trypanosomatidal sequences, an analysis was made of the conservation (or lack of conservation)

of catalytic and substrate binding-site residues in each domain For this purpose, we defined sets of key cata-lytic residues for each of the PFK-2 and FBPase-2 activities (boxed and highlighted in Fig 3): noncon-servative replacement of any of these residues would

be expected to abolish activity For PFK-2 activity key catalytic residues were Lys51, Thr52, Asp128 and Lys172 Site-directed mutation of these residues

Table 2 Properties of predicted PFK-2 ⁄ FBPase-2 isoenzymes of trypanosomatids.

Accession no Chromosome

No.

residues

Mass (kDa) a

Conservation

Presumed activity

PFK2-domain FBPase-2 domain Residues b

[key catalytic (total 4);

other binding (total 13)] Overall c (%)

Residues b

[key catalytic (total 5);

other binding (total 8)] Overall c (%)

Lm 1 LmjF3.0800 3 2422 251 4; 13 24–40 0; 2 20–23 PFK-2

Lm 2 LmjF26.0310 26 667 74 4; 10 17–29 3; 2 24–32 PFK-2

Lm 3 LmjF36.0150 36 485 55 1; 3 9–15 5; 8 27–36 FBPase-2

Lm 4 LmjF07.0760 7 1245 132 3; 6 13–25 4; 7 26–40 ? Tb1 Tb03.48O8.70 3 1023 111 4; 13 24–37 0; 2 18–23 PFK-2 Tb2 Tb07.27M11.980 7 648 72 2; 8 13–25 3; 2 23–29 ? Tb3 Tb10.70.2700 10 478 54 0; 4 10–16 5; 7 25–34 FBPase-2 Tb4 Tb08.29O4.60 8 702 79 2; 6 13–26 4; 8 30–37 ? Tc1–1 Tc00.1047053508153.950 1021 112 4; 13 23–41 0; 2 19–25 PFK-2 Tc1–2 Tc00.1047053508181.20 1023 113 4; 13 23–41 0; 2 10–24 PFK-2 Tc2–1 Tc00.1047053508207.230 705 79 3; 10 14–29 3; 2 24–30 ? Tc2–2 Tc00.1047053509509.30 702 79 3; 10 14–28 3; 2 24–29 ? Tc3–1 Tc00.1047053510963.50 481 55 0; 3 10–15 5; 8 28–36 FBPase-2 Tc3–2 Tc00.1047053508625.50 481 54 0; 3 10–15 5; 8 28–37 FBPase-2 Tc4–1 Tc00.1047053508569.130 749 84 2; 6 13–28 5; 7 28–40 FBPase-2 Tc4–2 Tc00.1047053503733.20 749 84 2; 6 13–27 5; 7 28–40 FBPase-2

a

Molecular mass calculated from ORF.bConservation of residues in the predicted trypanosomatid enzymes compared to functional mam-malian and S cerevisiae PFK-2 and FBPase-2 domains (and corresponding to boxed residues in Fig 3) c Overall percentage of amino acid sequence identity between the predicted trypanosomatid enzymes and functional PFK-2 and FBPase-2 domains from other eukaryotes.

Tb1 Tb2

Tb3 Tb4

500 residues

Fig 2 Diagrammatic representation of the domain structure of the

various PFK-2 ⁄ FBPase-2 isoenzymes of T brucei The domain

structure of the enzymes was inferred from the amino acid

sequences predicted from the ORFs Light grey, FBPase-2 domain;

dark grey, PFK-2 domain; white, insertions ⁄ extensions compared

with mammalian PFK-2 ⁄ FBPase-2 sequences.

B

Fig 3 Sequence alignments of the kinase domain (A) and bisphosphatase domain (B) In each case, the trypanosomatid sequences are compared with domains of confirmed activity, rat testis bifunctional enzyme (PDB code 2BIF) [24] in both cases, and the respective mono-functional S cerevisiae enzymes [SWISSPROT codes 6P21_YEAST in (A) and F26_YEAST in (B)] Numbers substitute large insertions to the rat testis enzyme For clarity, only one of each pair of T cruzi homologues is shown Rat testis enzyme numbering is shown beneath the alignment Key catalytic residues and additional binding residues are boxed, with the former also shown emboldened and italicized Within each box shading is used for functional conservation of the particular residue, i.e as an indication that the residue present would have the same capacity for electrostatic interaction, hydrogen bonding or hydrophobic interaction, as the residue present in rat testis enzyme The figures were produced with ALSCRIPT [88].

drastically reduces activity [33–35] Similarly, at the

FBPase-2 site, Arg255, His256, Arg305, Glu325 and

His390 were considered to be key residues [36–38] We

also considered other residues involved in substrate

binding (boxed in Fig 3) based mainly on X-ray

crys-tal structures The binding of ATP analogues in the

PFK-2 active site and of fructose 6-phosphate and

inorganic phosphate ions to the FBPase-2 active site

have been visualized crystallographically [24,25,39]

The binding site for fructose 6-phosphate in the

PFK-2 catalytic site has yet to be directly visualized,

but conserved residues in the vicinity have been the

subject of several docking and site-directed

mutagene-sis studies enabling modelling of fructose 6-phosphate

binding [26,35,40–42]

Striking differences in the patterns of conservation of

these residues were immediately apparent For

isoen-zyme 1, all key kinase catalytic residues were conserved

and additional substrate-binding residues were also well

conserved (Fig 3A; Table 2) In sharp contrast, no

FBPase-2 key catalytic residues were conserved

(Fig 3B; Table 2) For example, His256, which is

transi-ently phosphorylated during the catalytic cycle of

FBPase-2, is replaced by proline in all isoenzyme 1

sequences Comparisons of whole domains tell a similar

tale – the isoenzyme 1 kinase domains are 23–41%

iden-tical to presumed active PFK-2 domains, whereas the

corresponding range for the bisphosphatase domains is

18–25% Similarly, the isoenzyme 1 sequences are better

conserved in their N-terminal domains (59–68%

sequence identity, excluding the comparison of Tc1–1

and Tc1–2) than in their C-terminal domains (40–59%)

These data strongly suggest that Tb1, Lm1 and Tc1

homologues are monofunctional PFK-2s

Lm2 also conserves the key catalytic kinase residues

and none of the changes in other substrate binding

residues seems incompatible with kinase activity,

although Asn63 and Arg193, predicted to hydrogen

bond to fructose 6-phosphate, are replaced by Phe

and Val, respectively Surprisingly, the trypanosomal

representatives of isoenzyme 2 have nonconservative

replacements in the key kinase residues; Tb2 has Met

and Ala for Thr52 and Asp128, whereas both T cruzi

sequences replace Asp128 with Asn (Fig 3A; Table 2)

Given the key role of Asp128 in coordination to the

catalytically essential divalent metal cation [24], these

substitutions appear to rule out kinase activity for

Tb2, Tc1–1 and Tc1–2 Indeed, additional substrate

binding residues are less well conserved in these

sequences At the FBPase-2 site, both key catalytic

residues and additional substrate-binding residues are

poorly conserved (Fig 3B; Table 2) In particular, the

replacements of His256 and Glu325 rule out FBPase-2

activity for isoenzyme 2 Comparisons of whole domain conservation are uninformative: the kinase domains of the isoenzymes 2 are 13–29% identical to active kinase domains, whereas their bisphosphatase domains are 22–32% identical to other bisphosphatase domains Surprisingly, given the patterns of residue conservation, the C-terminal domain is slightly better conserved between isoenzyme 1 sequences (58–65%, again excluding the comparison of Tc2–1 and Tc2–2) than the N-terminal domain (47–58%) Taken together, the data suggest that Lm2 probably has monofunctional kinase activity but that experimental characterization would be needed to confirm the pre-diction Tb2, Tc1–1 and Tc1–2 are predicted, surpris-ingly, to have neither kinase nor FBPase-2 activity For isoenzyme 3, clear-cut predictions may once again be made In the kinase domain, key catalytic res-idues are absent from all sequences and other sub-strate-binding residues are not conserved (Fig 3A) At the FBPase-2 catalytic site, all residues are very highly conserved in all four sequences Similarly, the kinase domains of isoenzyme 3 sequences are just 9–15% identical with active domain sequences, whereas the bisphosphatase domain is well conserved at 25–36% identity Also, the C-terminal domain is much better conserved in an intertrypanosomatid comparison (62–73%) than the N-terminal domain (16–34%) Thus Lm3, Tb3, Tc3–1 and Tc3–2 would clearly be mono-functional FBPase-2 enzymes

In the kinase domain of isoenzyme 4, the replace-ment of Asp128 with Asn in all four sequences, as well

as various substitutions of Lys172, rules out kinase activity Additional substrate-binding residues are also poorly conserved (Fig 3A; Table 2) Isoenzyme 4 sequences are also much better conserved, compared with active domain homologues, in the bisphosphatase domain than in the kinase domain Conservation in the bisphosphatase domain is in the range 26–40% compared with just 13–28% in the kinase domain The corresponding figures for the intertrypanosomatid iso-enzyme 4 comparison are 47–59% for the bisphospha-tase domain and 34–53% identity in the kinase domain The T cruzi sequences have all the required FBPase-2 key catalytic residues and well-conserved additional substrate-binding residues The other two isoenzyme 4 sequences have nonconservative replace-ments of key catalytic residues; the substitution of Arg255 by Leu in Lm4 and the replacement of His256

by Asn in Tb4 These argue against their having FBPase-2 activity, but in each case mutations else-where in the catalytic site make it difficult to com-pletely rule out activity The loss of Arg255, a residue that binds the 2-phospho group of substrate and the

phosphohistidine intermediate [37,43], could be

parti-ally compensated by the presence of neighboring

His416 (replacing Leu in rat testis) which could form

an ionic interaction with the 2-phosphate group

Equally, the lack of phosphorylable His256 in Tb4

would typically be thought to be sufficient to abolish

bisphosphatase activity However, experiments on

FBPase-2 and relatives show that caution should be

exercised Most importantly, when this His was

replaced by Ala in the rat testis enzyme, a surprising

17% of catalytic activity was maintained [44], probably

due to water taking over the nucleophilic role [39] It

is also relevant to note the surprising variations in key

catalytic residues in members of the related

phospho-glycerate mutase superfamily [45] In the case of Tb4,

it is also intriguing to note the replacement of Asn262

(in rat testis enzyme) with an additional His at

posi-tion 262 The new His is well placed to interact with

the 2-phosphate group of the incoming substrate

However, it seems unlikely that this new His may take

over the role of the missing His256 to form the

phos-pho-enzyme intermediate, as it is not suitably

posi-tioned for in-line attack on the 2-phosphate group of

the substrate In summary, although Tc4–1 and Tc4–2

are probably monofunctional bisphosphatases,

experi-mental data would be required to test the possibility of

Lm4 and Tb4 sharing the same function

Heterologous expression, purification and

characterization of T brucei PFK-2/FBPase-2

We set out to test experimentally the results of the

bio-informatics analysis of sequences retrieved from the

databases To that end, PCR amplification was first

performed for Tb1 and Tb4 fragments, using as

template genomic DNA from our laboratory strain,

T bruceistock 427 The fragments thus obtained were

used to screen an available genomic library DNA

fragments were subcloned in plasmids and sequenced

All clones obtained contained either of the two distinct

genes The Tb1 and Tb4 sequences in the database of

T brucei stock TREU927⁄ 4 have the same length as

the proteins encoded by the genes analysed by us for

T brucei stock 427 (Table 2), but differ by a number

of substitutions For Tb1, substitutions were found at

five positions (Thr18Met, Pro51Leu, His384Arg,

Ala422Gly and Lys630Glu) Only the latter

substitu-tion, corresponding to position 138 in the rat testis

enzyme [24], is within the PFK-2 domain, but the

resi-due is not part of the active site The other four

posi-tions lie in the N-terminal extension The Tb4 amino

acid sequences of stocks TREU927⁄ 4 and 427 differ

at 12 positions by single amino acid changes; two

substitutions (Ser–Asn at position 536 of the full-length predicted protein and His–Gln at position 589) are within the PFK-2⁄ FBPase-2-specific region, but are not expected to have any consequence for enzyme activity The first of these positions is on an insertion relative to the rat testis enzyme and the second corres-ponds to position 347 These differences should, most likely, be attributed to polymorphisms between the

T bruceistrains used by us and in the genome-sequen-cing project

For Tb2, the full-length gene of stock 427 was amplified and sequenced No differences were found between the Tb2 nucleotide sequences of the two stocks

To prove the identity of Tb1 as an active PFK-2, and to confirm that the Tb2 is inactive as predicted,

we expressed the proteins in a heterologous system However, it was anticipated that the large Tb1 poly-peptide as predicted from the full-length open reading frame (ORF), would be difficult to express as a soluble active protein Therefore, a shorter part comprising the region homologous to the bifunctional

PFK-2⁄ FBPase-2 enzymes of higher eukaryotes was chosen for expression A protein starting at codon ATG 505 (giving the N-terminal sequence MSSSYTTVSDAVSL-) corresponds quite well with the beginning of the struc-turally resolved part of the rat testis bifunctional enzyme and from where good alignment is possible (the multiple alignment in Fig 3A starts with the underlined last three residues) Moreover, this protein still contains the region corresponding to the N-ter-minal part of other PFK-2s that is involved in dimeri-zation The shorter ORF codes for a polypeptide of

519 amino acids (including the initiator methionine), with a calculated molecular mass of 57 078 Da and a

pI value of 9.29 When expressed with a His-tag, as described in Experimental procedures, a protein of 547 residues with a predicted molecular mass of 60 306 Da and a pI of 9.29 is produced Under rather specific growth conditions, adapted from Oza et al [46], low amounts of soluble enzyme could be obtained that indeed displayed PFK-2 activity The protein was par-tially purified On SDS⁄ PAGE several bands were visible, but the identity of a polypeptide of Mr

60 000 was confirmed as Tb1 after western blotting and immunodetection with anti-(poly His) sera (not shown) Expression of larger constructs was also attempted, both using Escherichia coli cells and

in vitro, in a coupled transcription–translation system (Rapid Translation System, Roche Molecular Bio-chemicals), using different vector systems, differently placed tags for affinity purification, and a variety of conditions for bacterial growth and induction of

protein expression However, in most cases very poor

expression of an inactive, unstable protein was

obtained, or the protein was expressed as insoluble

inactive enzyme

The active 60 000 MrTb1 was subjected to a

prelim-inary kinetic analysis The K app

m for fructose 6-phos-phate varied between 1.9 ± 0.12 and 4.6 ± 0.8 mm,

whereas the K app

m for ATP was between 1.6 ± 0.27

and 2.0 ± 0.30 mm in four different preparations of

enzyme These values are similar to those for the

enzyme partially purified from bloodstream-form

try-panosomes (see above) and to values reported

previ-ously [4] With regard to their PFK-2 activity, the

various mammalian isoenzymes display much lower

Km values: 15–150 times for fructose 6-phosphate and

3–20 times for ATP [4,16]

The relatively good conservation of the ATP-binding

site residues between mammalian and Tb1 proteins

suggests that the explanation for the lower ATP

affin-ity of the latter must lie with the three significantly

dif-ferent positions, residue 220 (Val in mammalian

enzymes, Lm1 and Tc1, but Ala in Tb1), residue 246

(Val or Ile in mammalian enzymes, Pro in the

trypan-osomatid proteins) and position 427 (Tyr in

mamma-lian enzymes, Gly, Glu and Asp in Lm1, Tc1 and Tb1,

respectively) The branched side chains of residues 220

and 246 form the side of the adenine-binding pocket

further away from the catalytic site and each make

multiple hydrophobic interactions with the heterocyclic

ring Their replacement with nonbranched Ala and Pro

would reduce the steric complementarity of adenine

and its pocket, thereby reducing the strength of the

interaction Also, the hydrogen bond from Tyr427 to

the a-phosphate of ATP is not present in Tb1 Instead,

the replacement Asp could, assuming local correctness

of the sequence alignment, lead to electrostatic

repul-sion of the negatively charged phosphate groups of

ATP

Bacterially expressed Tb1 was analysed by gel

filtra-tion over a Superdex 200 HR 10⁄ 30 column, to

deter-mine its oligomeric state However, under all

conditions tested (various buffers, variable ionic

strength, presence of reducing agents – Experimental

procedures) all Tb1, as detected by western blotting

using an antiserum specific for the His6-tag, eluted as

an entity of high mass (> 600 kDa) with a low PFK-2

activity In addition, a second protein peak with higher

PFK-2 activity eluted with a relative molecular mass

of
140 kDa, suggesting it was a dimer Purified rat

liver PFK-2⁄ FBPase-2, used as a control, eluted as a

110 kDa dimer as detected by both PFK-2 activity and

western blotting using a homologous antiserum These

results suggest that the bacterially produced Tb1

enzyme is an active dimer that has a strong tendency

to aggregate

In contrast to Tb1, no activity was found for bacte-rially expressed, soluble Tb2, in agreement with the predictions of the sequence analysis Surprising in this respect is that the Tb2 gene sequences found in the two different T brucei stocks were identical and that

an Expressed Sequence Tag corresponding to Tb2 has been found in procyclic T brucei rhodesiense libraries (GenBank accession no AA689209.1) This suggests that this protein, despite its lack of PFK-2 activity, may play a role in these trypanosomes

The fact that only Tb1 displays activity, and that no other sequences with typical PFK-2 features could be detected in any of the trypanosomatid databases, strongly suggests that the 76.4 kDa protein purified from trypanosomes is a Tb1 form, despite the fact that the complete ORF of Tb1 gene codes for a 110 kDa polypeptide We hypothesize, therefore, that the puri-fied protein represents a processed form of the protein (see also below) Future studies will also include a detailed experimental analysis of the two T brucei pro-teins with likely and possible FBPase-2 activity, Tb3 and Tb4, respectively A Tb4 construct of stock 427 trypanosomes has already been expressed in E coli, although so far in mainly an insoluble form, and Tb3 will be expressed in the future FBPase-2 assays were not performed for Tb1 and Tb2, as it is inconceivable that these proteins would possess any FBPase-2 activ-ity, as explained above (also Table 2)

Evolution of PFK-2/FBPase-2 The amino acid sequences of the bifunctional PFK-2⁄ FBPase-2 enzymes from many organisms were retrieved from the swissprot database, aligned with those of the various trypanosomatid PFK-2⁄ FBPase-2 homologues and used for evolutionary analysis All the PFK-2⁄ FBPase-2 sequences can be conveniently divi-ded into an N-terminal PFK-2 domain and a C-ter-minal FBPase-2 domain In the yeast PFK27 sequence, the FBPase-2 domain is difficult to recognize Presum-ably its sequence diverged considerPresum-ably and was trun-cated during evolution The PFK-2 domain is related

to a family of nucleotide-binding proteins including adenylate kinase, p21 ras, EF-Tu, the mitochondrial ATPase b-subunit and myosin ATPase, all of which have a similar fold and contain the Walker A and B motifs The FBPase-2 domain also belongs to a protein family comprising the cofactor-dependent phosphogly-cerate mutases and acid phosphatases The bifunc-tional PFK-2⁄ FBPase-2 must have originated by fusion

of representatives of these two families in a common

ancestor of all eukaryotic organisms studied here

(try-panosomatids, yeasts and fungi, plants and animals)

The PFK-2 and FBPase-2 domains can be flanked by

extensions of variable lengths In plants, Neurospora

crassaand yeast PFK26 the N-terminal extensions can

be long and increase the molecular mass from
55 to

90 kDa From the ORFs, we infer that in some

try-panosomatid isoenzymes, the N-terminal extensions

can be even longer The extremities of mammalian

iso-enzymes serve as regulatory domains often containing

phosphorylation sites Moreover, it has been shown

that the N-terminal domain of the Arabidopsis enzyme

is important both for subunit assembly and for

defi-ning the kinetic properties of the enzyme [47] In each

of the yeast and trypanosomatid isoenzymes, one of

the catalytic cores seems to have been inactivated,

ren-dering the bifunctional enzyme monofunctional

The sequences of PFK-2⁄ FBPase-2 in animals and

plants form distinct clusters in phylogenetic trees made

separately for the PFK-2 and FBPase-2 domains, with

adenylate kinase and phosphoglycerate mutase as

out-groups, respectively (Fig 4) In plants, only a single

gene of the bifunctional enzyme was detected [18,48],

whereas animals have different bifunctional isoen-zymes, represented by four subtrees (corresponding to the liver⁄ muscle, heart⁄ kidney, testis and brain⁄ placenta groups, respectively) The evolution of the PFK-2⁄ FBPase-2 in the lineages of yeasts ⁄ fungi and Trypanosomatidae is more difficult to deduce from the phylogenetic trees This is due to: (a) the relatively low level of conservation of these sequences, and the longer evolutionary distances when these organisms are con-sidered; and (b) the fact that the inactivated domains

of these enzymes have possibly been subject to a very high evolution rate For the same reason, the highly aberrant C-terminal domain of S cerevisiae PFK27 was omitted from the FBPase-2 tree Nevertheless, our preliminary analysis suggests that most isoenzymes of the fungi result from gene duplications within this group Furthermore, in the FBPase-2 domain tree, all the Trypanosomatidae sequences are together in one cluster separated from the sequences of all other organisms (but containing the outgroup) The iso-enzymes form individual groups Isoiso-enzymes 1 and 2, containing all the putative PFK-2s (Lm1, Tb1, Tc1–1, Tc1–2 and Lm2), cluster together, as do isoenzymes 3

Fig 4 Phylogenetic trees of the PFK-2 and FBPase-2 domains of both the bifunctional and monofunctional proteins All PFK-2 and ⁄ or FBPase-2 containing sequences from animals, invertebrates, fungi and protists, as obtained from the SWISSPROT ⁄ TREMBL databases (Experimental procedures) were aligned with each other using CLUSTALX [80] From this alignment subalignments were created containing either the PFK-2 domain or the FBPase-2 domain Each of the subalignments was used for the creation of a neighbour-joining tree from a matrix of uncorrected pair-wise distances using the tree option within CLUSTALX Regions with insertions or deletions were excluded from the analyses Horizontal bars represent 10 substitutions per 100 residues The trees were rooted using either an Arabidopsis thaliana chloroplast adenylate kinase or E coli cofactor-dependent phosphoglycerate mutase as an outgroup.

and 4, containing all the likely FBPase-2s (Lm3, Tb3,

Tc3–1, Tc3–2, Tc4–1 and Tc4–2) The situation for the

trypanosomatids in the PFK-2 tree is very similar,

except also for the presence of the S cerevisiae PFK27

in the cluster Furthermore, the phylogenetic analysis

also showed that the formation of the four isoenzymes

in the trypanosomatids has occurred already in the

common ancestor of the genera Trypanosoma and

Leishmania

The presence of multiple isoforms of PFK-2⁄

FBPase-2 in mammals can be understood as a need

for distinct enzymes each with different kinetic

proper-ties and regulatory mechanisms optimized in regulating

glycolysis and⁄ or gluconeogenesis in the different

tis-sues [1,16] With regard to yeast, the two isoenzymes

with PFK-2 activity differ in that only PFK26 is

acti-vated by protein phosphorylation [29], whereas the

synthesis of PFK27 is only induced by fermentable

carbon sources [21] However, the growth rates and

glycolytic flux of both the PFK26 and PFK27 deletion

mutants of S cerevisiae are similar to that of wild-type

cells [19,21], and did not reveal an essential role of

Fru2,6-P2 in the regulation of carbon fluxes in this

organism [49] Nor could different roles for the two

PFK-2s be demonstrated by metabolome analysis of

the mutants [50]

Our data do not permit us to draw any conclusions

as to the reason why trypanosomatids have four

iso-enzymes Sequence inspection and structure modelling

suggested that some of them (Tb1, Lm1, Tc1–1,

Tc1–2 and Lm2) are monofunctional PFK-2s,

whereas others (Tb3, Lm3, Tc3–1, Tc3–2, Tc4–1 and

Tc4–2) most likely only have FBPase-2 activity The

sequence analysis suggested that these proteins are all

soluble It did not reveal obvious topogenic signals

indicative for functioning of isoenzymes in different

cell compartments Only Tb2, the inactive PFK-2 of

T brucei, contains a potential peroxisome-targeting

signal at its C-terminus (-NKL) [51], but a similar

tripeptide was not found on the corresponding

sequences of the other trypanosomatids It could be

imagined that isoenzymes with different properties are

necessary at different stages of the life cycle of these

organisms Many trypanosomatid species are

patho-genic organisms with a highly complicated life cycle

T brucei cycles between the mammalian bloodstream,

the tsetse fly midgut and the insect’s salivary gland

These are radically different environments where the

parasite encounters different nutrients and has to

adapt its metabolism accordingly Leishmania species

undergo similar transitions between flies and

mam-mals where they predominantly live intracellularly in

the phagolysosomes of macrophages

Why the bifunctional PFK-2⁄ FBPase-2 evolved into different monofunctional enzymes in the yeasts and, most likely, also in trypanosomatids is not clear Poss-ibly, it represents an as yet not understood adaptation

to the specific requirements of glucose metabolism in these unicellular organisms, different from the require-ments in the ancestral eukaryote where the fusion of the PFK-2 and FBPase-2 domains in a single enzyme occurred and different from that in extant animals and plants with their bifunctional enzymes Usually, oppos-ite metabolic reactions are catalysed by separate enzymes; PFK-2⁄ FBPase-2 is an exception Moreover, bifunctionality extends to its substrate⁄ product, Fru2,6-P2, which in higher eukaryotes (but not in try-panosomatids) has two targets, PFK-1 and FBPase-1 The advantages of the association of opposite reac-tions may be: (a) simplicity in short-term control, such

as regulation at a single site by an allosteric effector or phosphorylation; and (b) simplicity of long-term regu-lation (one gene, one mRNA) One possible reason why several microorganisms have, at a later stage of evolution, again uncoupled the PFK-2 and FBPase-2 activities is that it may have endowed an increased flexibility to adapt to different growth conditions: dif-ferent combinations of monofunctional PFK-2 and FBPase-2 enzymes may be expressed in cells growing

in different environments However, this remains to be studied by following the expression of the different enzymes throughout the life cycle of the trypanosomat-ids That each of the monofunctional enzymes retained (part of) the inactivated domain is in line with the notion that both domains may be required for proper folding or oligomerization

The inferred activities of some trypanosomatid iso-enzymes remain to be confirmed Enzymes with similar activities may differ in kinetic and regulatory proper-ties We have demonstrated that Tb1 has PFK-2 acti-vity, whereas no activity could be found for Tb2, in agreement with sequence analysis predictions The apparent Km values of Tb1 for fructose 6-phosphate and ATP are similar to those of the enzyme partially purified from the bloodstream-form trypanosomes (Table 1) [4] The lack of FBPase-2 activity in this enzyme is highly likely It should be noted, however, that we do not know if the absence of N-terminal domains may have affected the activity of the bacteri-ally expressed enzyme

It is interesting to note that Expressed Sequence Tags corresponding to Tb2 (GenBank accession no AA689209.1; unpublished data) and Tb3 (T26149) [52] have been obtained from procyclic T brucei rhodes-ienselibraries Our unpublished data (N Chevalier and

P A M Michels, unpublished) showing the presence