Báo cáo khoa học: Characterization of the cofactor-independent phosphoglycerate mutase from Leishmania mexicana mexicana docx

The resulting solutions were tested for mutase activity in the presence of different potentially activating or inhibiting compounds, in order to characterize the type of mutase present i

Characterization of the cofactor-independent phosphoglycerate

Histidines that coordinate the two metal ions in the active site show different

susceptibilities to irreversible chemical modification

Daniel G Guerra1, Didier Vertommen2, Linda A Fothergill-Gilmore3, Fred R Opperdoes1

and Paul A M Michels1

1

Research Unit for Tropical Diseases, and2Hormone and Metabolic Research Unit, Christian de Duve Institute of Cellular Pathology and Laboratory of Biochemistry, Universite´ Catholique de Louvain, Brussels, Belgium;3Structural Biochemistry Group,

Institute of Cell and Molecular Biology, University of Edinburgh, UK

Phosphoglycerate mutase (PGAM) activity in promastigotes

of the protozoan parasite Leishmania mexicana is found only

in the cytosol It corresponds to a cofactor-independent

PGAM as it is not stimulated by 2,3-bisphosphoglycerate

and is susceptible to EDTA and resistant to vanadate

We have cloned and sequenced the gene and developed a

convenient bacterial expression system and a high-yield

purification protocol Kinetic properties of the bacterially

produced protein have been determined

(3-phosphoglycer-ate: Km¼ 0.27 ± 0.02 mM, kcat¼ 434 ± 54 s)1;

2-phos-phoglycerate: Km¼ 0.11 ± 0.03 mM, kcat¼ 199 ± 24 s)1)

The activity is inhibited by phosphate but is resistant to Cl–

and SO42– Inactivation by EDTA is almost fully reversed by

incubation with CoCl2but not with MnCl2, FeSO4, CuSO4,

NiCl2or ZnCl2 Alkylation by diethyl pyrocarbonate

resul-ted in irreversible inhibition, but saturating concentrations of

substrate provided full protection Kinetics of the inhibitory reaction showed the modification of a new group of essential residues only after removal of metal ions by EDTA The modified residues were identified by MS analysis of peptides generated by trypsin digestion Two substrate-protected histidines in the proximity of the active site were identified (His136, His467) and, unexpectedly, also a distant one (His160), suggesting a conformational change in its envi-ronment Partial protection of His467 was observed by the addition of 25 lMCoCl2to the EDTA treated enzyme but not of 125 lM MnCl2, suggesting that the latter metal ion cannot be accommodated in the active site of LeishmaniaPGAM

Keywords: chemical modification; kinetics; Leishmania mexicana; metal dependence; phosphoglycerate mutase

The reversible isomerization of 2-phosphoglycerate (2PGA)

and 3-phosphoglycerate (3PGA) is an obligate step for both

glycolysis and gluconeogenesis This step is carried out

in two different ways in nature, by two different types

of evolutionarily unrelated enzymes (although both

EC 5.4.2.1) The better documented enzyme is the cofac-tor-dependent phosphoglycerate mutase (d-PGAM) due to its requirement for 2,3-bisphophoglycerate It is present in some eubacteria, yeast and all vertebrates most frequently as

a dimer or tetramer of 23–30-kDa subunits [1] The second enzyme, called cofactor-independent phosphoglycerate mutase (i-PGAM), is a monomeric protein of 60 kDa Upon comparative sequence and structure analysis, the former enzyme has been classified as a member of the phospho-histidine acid phosphatase superfamily [2] and the latter as a member of the metal-dependent alkaline phosphatase superfamily [3,4] Whereas d-PGAM is the enzyme present in all vertebrates, i-PGAM is found in all plants and archaebacteria [5] and, together with d-PGAMs,

in lower eukaryotes and eubacteria [1,6]

We have previously shown that an i-PGAM participates

in glycolysis in the protist Trypanosoma brucei [7], a human pathogen The completely distinct structures and catalytic mechanisms of trypanosomal and human PGAM offer great promise for the design of inhibitors with high selectivity for the parasite’s enzyme Therefore, this finding should aid in the search for new drugs that are needed against diseases caused by members of the trypanosomatid family (Trypanosoma, Leishmania) [8–11] for which glucose catabolism is of vital importance

Correspondence to P A M Michels, ICP-TROP 74.39, Avenue

Hippocrate 74, B-1200 Brussels, Belgium Fax: + 32 27626853,

Tel.: + 32 27647473, E-mail: michels@bchm.ucl.ac.be

Abbreviations: GAPDH, glyceraldehyde-3-phosphate dehydrogenase;

DEPC, diethyl pyrocarbonate; ENO, enolase; LDH, lactate

dehydrogenase; PEP, phosphoenolpyruvate; PGA, phosphoglycerate;

d-PGAM, cofactor-dependent phosphoglycerate mutase; i-PGAM,

cofactor-independent phosphoglycerate mutase; PGK,

phospho-glycerate kinase; PYK, pyruvate kinase; TEA, triethanolamine.

Enzymes: glyceraldehyde-3-phosphate dehydrogenase (EC 1.2.1.12);

enolase/2-phospho- D -glycerate hydrolase (EC 4.2.1.11); lactate

dehy-drogenase (EC 1.1.1.27); phosphoglycerate kinase (EC 2.7.2.3);

phos-phoglycerate mutase (EC 5.4.2.1); pyruvate kinase (EC 2.7.1.40).

Note: The novel nucleotide sequence data published here have been

deposited in the EMBL-EBI/GenBank and DDBJ databases and are

available under accession number AJ544274.

(Received 20 January 2004, revised 25 February 2004,

accepted 19 March 2004)

In the study reported here we measured a PGAM activity

in lysates of cultured promastigotes (representative of the

insect-infective stage) of Leishmania mexicana, identified

it as cofactor-independent and located it to the cytosol of

the parasite We have cloned the gene of this enzyme

(LmPGAM) from a genomic library, expressed it to

produce an active His-tagged protein in Escherichia coli

and purified the enzyme with a single metal affinity column

The availability of this convenient expression and

purifica-tion system allowed us to undertake high-resolupurifica-tion

crys-tallographic studies [12] We also present a detailed

biochemical characterization of the bacterially produced

enzyme and its dependency on metal ions We studied the

accessibility of different residues involved in interactions of

the enzyme with the substrate and metal ions via chemical

modification combined with MS analysis Our results on

irreversible inhibition strongly suggest that the design of a

substrate analogue as an irreversible inhibitor is feasible and

pave the way for the development of selective inhibitors that

may be used as lead compounds for trypanocidal drugs

Experimental procedures

Growth, harvesting and fractionation of parasites

Promastigotes of L mexicana mexicana strain NHOM/B2/

84/BEL46 were grown at 28C in the semidefined medium

SDM-79 [13], supplemented with 10% (v/v)

heat-inacti-vated foetal bovine serum (Gibco) After 4 days, cells in the

exponential phase of growth (8.7· 107 cellsÆmL)1) were

harvested by centrifugation, washed twice in an iso-osmotic

buffer containing 3 mM imidazole (pH 7.0) and 250 mM

sucrose, and immediately lysed by mixing to a thick paste

with silicon carbide powder previously washed with ethanol

and water, and grinding The lysate was cleared by

centrifugation at 30 g and different cell fractions were

obtained by subsequent centrifugation steps at 1500 g;

cellular extract [S3.5] and nuclear fraction [P3.5], 5000 g;

large-granular fraction [P6.5], 15 000 g, small-granular

fraction [P11] and 140 000 g, microsomal fraction [P40]

and cytosolic fraction [S40] As described previously [14], all

procedures were performed at 4C

Enzyme assays

PGAM activity was measured by following either the

increase of UV absorbance at 240 nm due to

phosphoenol-pyruvate (PEP) production (molar extinction coefficient

1310M )1Æcm)1) or the decrease of UV absorbance at 340 nm

due to NADH oxidation (molar extinction coefficient

6250M )1Æcm)1) using a Beckman DU7 spectrophotometer

NADH oxidation, forward reaction The conversion of

3PGA to 2PGA was coupled to NADH oxidation by

lactate dehydrogenase (LDH) via enolase (ENO) and

pyruvate kinase (PYK), and following the concomitant

decrease of absorbance at 340 nm The assay was

per-formed at 25C in a 1-mL reaction mixture containing

0.1M triethanolamine (TEA)/HCl pH 7.6, 1 mM MgCl2,

1 mM ADP, 0.56 mM NADH, 0.1 mM CoCl2, 1.5 mM

3PGA, and the auxiliary enzymes ENO, PYK and LDH

at final activities of 0.55, 8.0 and 13.8 UÆmL)1, respectively

NADH oxidation, reverse reaction The conversion of 2PGA to 3PGA was coupled to NADH oxidation by glyceraldehyde-3-phosphate dehydrogenase (GAPDH) via 3-phosphoglycerate kinase (PGK) The assay was per-formed at 25C in a 1-mL reaction mixture containing 0.1MTEA/HCl pH 7.6, 5 mMMgCl2, 1 mMdithiothreitol,

1 mM ATP, 0.56 mM NADH, 0.01 mM CoCl2, 0.8 mM

2PGA, and GAPDH and PGK both at 6 UÆmL)1 The CoCl2added in the reverse reaction assay was lower in order

to avoid the formation of pink precipitates of cobalt in the presence of dithiothreitol

PEP production, forward reaction The reaction was followed upon the addition of 1.5 mM3PGA into a 1 mL quartz cuvette containing 50 mM Hepes pH 7.6, 0.55 U

of rabbit muscle ENO, 1 mMMgCl2, 0.1 mM CoCl2 and

50 mM KCl One activity unit (U) is defined as the conversion of 1 lmol substrateÆmin)1 under standard conditions

Auxiliary enzymes used were obtained from Roche Molecular Biochemicals (rabbit muscle PYK and GAPDH and yeast PGK) and Sigma (rabbit muscle ENO and bovine heart LDH)

Measuring PGAM activity in aLeishmania lysate The mutase activity was measured in 50 lL of the lysate and different subcellular fractions by following PEP production

as described above Measurements were repeated after an overnight dialysis of all fractions at 4C against 200 vols of 0.1M Hepes pH 7.6, 0.5M NaCl, 25 mM imidazole and 0.1 mM CoCl2, in order to remove potentially interfering metabolites The resulting solutions were tested for mutase activity in the presence of different potentially activating or inhibiting compounds, in order to characterize the type

of mutase present in Leishmania: 50 lM NaVO3, 0.6 mM

2,3-bisphosphoglycerate (Sigma) and after incubation with

5 mMEDTA In parallel, the bacterially produced, purified

L mexicanai-PGAM (see below) and commercially avail-able rabbit muscle d-PGAM (Roche Molecular Biochem-icals) were also assayed in the presence of these compounds Library screening, subcloning and sequencing

The T brucei i-PGAM gene [7] was used as a template to obtain a 693 bp PCR product corresponding to a generally well conserved part of i-PGAMs (corresponding to residues 71–302 in Fig 2) The amplified DNA was purified after electrophoresis through agarose, labelled with32P by nick translation and used as a hybridization probe against blots

of 10 plates of E coli infected with approximately 4000 plaque forming units per plate (approximately 15 times the genome size) of a genomic library of L m mexicana prepared in the phage vector kGEM11 [15] Double digestion of DNA purified from a positive phage k clone with SacI and HindIII restriction enzymes yielded a 6 kb fragment that hybridized with the probe The 6 kb fragment was ligated into plasmid pZErO-2 (Invitrogen) and used to transform E coli XL1-blue cells Further digestion of the plasmid with EcoRI gave a positive band of 3.7 kb which was analysed by automatic sequencing using a Beckman CEQ 2000 sequencer

Sequence analysis

ABLASTPquery was performed with the newly determined

L mexicanaPGAM amino-acid sequence (LmPGAM) in

the EMBL-EBI site (http://www.ebi.ac.uk/blast2/) against

the SwissProt database All i-PGAM sequences recognized

were retrieved and stored locally Sequences corresponding

to T brucei, Bacillus stearothermophilus and Caenorhabditis

eleganswere also appended for a multiple alignment using

the program CLUSTALX (BLOSUM matrix series, default

settings) Uncorrected distances between the i-PGAM

sequences belonging to archaebacteria and all other

sequences showed values higher than 0.85 and therefore

this group was not included in any further analysis despite

their proven i-PGAM activity [5,16] A bootstrapped

unrooted neighbour-joining tree was created with the

remaining amino acid sequences, ignoring positions with

gaps in the alignment

An automatic alignment performed by SwissPdbViewer

between the amino acid sequences of B

stearothermo-philusand L mexicana was corrected manually using the

information from theCLUSTALX multiple alignment Then

the L mexicana sequence was threaded into the structures

of the B stearothermophilus enzyme cocrystallized with

2PGA and 3PGA (PDB codes: 1EQJ and 1EJJ) The

positions of important amino acids were confirmed by

examining every residue within a 7 A˚ radius of the 3PGA

substrate bound in the active site

Construction of a bacterial expression system

The entire LmPGAM gene was amplified by PCR with the

proofreading Vent DNA polymerase (New England

Bio-labs) while adding restriction sites for NcoI and XhoI at both

of its flanks The PCR product was treated with Taq DNA

polymerase for the addition of overhanging A nucleotides to

enable insertion into the pGEM-T easy vector (Promega) by

annealing of cohesive ends Transformed E coli XL1-blue

cells were plated on Luria–Bertani agar with ampicillin as

selective antibiotic The insert with the L mexicana gene

was excised from the plasmid by double digestion with NcoI

and XhoI and ligated into a similarly treated plasmid

pET28a The resulting plasmid pET28LmPGAM was used

to transform E coli XL-1 blue and BL21 cells; kanamycin

was used as antibiotic for selection of recombinant clones

The sequence appeared to be identical to the genomic

sequence determined earlier, except for a single nucleotide

difference resulting in a SfiA substitution of the second

amino acid as a consequence of the creation of the NcoI

restriction site and the presence of a tag at the C terminus

(translated as LEHHHHHH) The C-terminally tagged

protein thus produced is called C-LmPGAM It should be

noted that a PCR fragment amplified from a genomic clone

could be used for insertion in the expression vector, because

the Leishmania i-PGAM gene does not contain any intron

The absence of introns in protein-coding genes is a general

feature of trypanosomatids

Production, purification and storage of protein

Optimal growth conditions were standardized in order to

obtain the highest amount of total soluble protein, as

assessed by the intensity of a 60 kDa band on SDS/ PAGE and by total mutase activity E coli BL21 cells harbouring the pET28aLmPGAM recombinant plasmid were grown at 37C in 50 mL Luria–Bertani medium with 30 lgÆmL)1 kanamycin for approximately 4 h until the culture reached D600 of 0.5–0.7 Production of the C-LmPGAM was then induced by adding isopropyl thio-b-D-galactoside at a final concentration of 1 mM, and the culture was transferred to a water bath at 17C After continued growth with agitation for approximately 20 h, the cells were harvested by centrifugation and stored at )20 C

Cell pellets were resuspended in 5 mL ice cold lysis– equilibration buffer containing 0.1M TEA/HCl pH 8.0, 0.5M NaCl, 10% (v/v) glycerol and a protease inhibitor mixture (Roche Molecular Biochemicals) and broken by two passages through a French pressure cell at 90 MPa Approximately 10 mg of protamine sulphate was mixed with the lysate that was subsequently centrifuged (10 000 g, 20 min, 4C) Virtually all mutase activity measured in the supernatant was vanadate resistant and therefore due to i-PGAM The supernatant was then passed through 1.5 mL of TALON (Clontech) resin packed in a column connected to a peristaltic pump Fractions of  0.9 mL were collected and the protein content was estimated by measuring absorption of UV light at 280 nm The column was washed with equilibra-tion buffer and with a stepwise gradient of imidazole using concentrations of 5, 10 and 25 mM Fractions corresponding to protein peaks were examined by SDS/ PAGE followed by Coomassie blue staining A major band of approximately 60 kDa appeared at 10 and

25 mM imidazole fractions, and were pooled separately for further assays

To determine optimal storage conditions of C-LmPGAM, its specific activity as measured shortly after TALON purification [protein in 0.1M TEA pH 8, 10% (v/v) glycerol, 0.5M NaCl, 25 mM imidazole and 0.1 mM

CoCl2] was compared with that after incubation at different conditions To that purpose, 4 mL of the purified protein was concentrated approximately fourfold using a Centricon centrifugal filter unit (Millipore) and subsequently desalted

by passing through a 5 mL Sephadex G-25 column equilibrated with 0.1MTEA pH 7.6 Fractions of 0.5 mL were taken and the absorbance at 280 nm and conductivity measured to assess their content of protein, imidazole and NaCl, respectively The protein peak fractions were collec-ted; the enzyme specific activity was checked and it appeared essentially the same as before the treatment The desalted protein was then diluted 1 : 1 into five different buffers

of the following final composition: A, 0.1MTEA pH 7.6;

B, 0.1MTEA pH 7.6, 0.5MNaCl; C, 0.1MTEA pH 7.6, 0.5M NaCl, 20% (v/v) glycerol; D, 0.1M TEA pH 7.6, 0.5M NaCl, 20% (v/v) glycerol, 0.1 mMCoCl2; E, 0.1M

TEA pH 7.6, 0.5M NaCl, 20% (v/v) glycerol, 0.1 mM

CoCl2, 25 mM imidazole; the protein concentration was 0.10 mgÆmL)1 for all five conditions The mixtures were incubated at 4C, and the stability of the protein under each condition was followed in time by regularly measuring the activity

Protein concentrations were measured by the Bradford assay [17], using BSA as standard

Determination of kinetic parameters (Km,kcat)

and pH optima

Forward reaction To determine accurately the kinetic

constants, the concentration of 3PGA was measured

enzymatically just before the assays For Km calculation,

15 assays were performed spanning a range of different

concentrations of 3PGA from 0.09 to 4.53 mM The

resulting data were fitted by a hyperbolic curve according

to the Michaelis–Menten equation and evaluated by the

minimal-squares method Evaluation was also done by

preparing linear plots according to Lineweaver–Burk (linear

regression coefficient, r2¼ 0.9975), Hanes (r2¼ 0.9986)

and Eadie–Hofstee (r2¼ 0.9775)

Reverse reaction For Km calculation, seven assays were

performed spanning different concentrations of 2PGA

ranging from 0.03 to 0.64 mM The resulting data were

similarly fitted by a Michaelis–Menten curve and linearized

plots [Lineweaver–Burk (r2¼ 0.9965), Hanes (r2¼ 0.9995)

and Eadie–Hofstee (r2¼ 0.9896)]

Reactivation by metals

In order to determine the time and pH dependency of

the inactivation by EDTA, aliquots of stored LmPGAM

were diluted 1 : 1 at a protein concentration of 0.2–

0.35 mgÆmL)1in an appropriate buffer (Mes or Hepes) to

reach pH 6.2, 7.0 or 8.0, as confirmed by indicator paper,

and incubated overnight at 4C Specific activity was then

measured and found similar for each sample Subsequently,

EDTA was added to a final concentration of 1 mMand the

incubation continued for different periods of time The

incubation was stopped by diluting an aliquot 25 times in

Hepes pH 7.6, 1 mMdithiothreitol, 0.1 mgÆmL)1BSA and

measuring its activity The reactivation of the

EDTA-treated protein was tested by adding metal ions to the final

sample taken (pH 8.0, 6h 15 min) This was done by

addition of 100 lMof either MnCl2, FeSO4, CoCl2, NiCl2,

CuSO4or ZnCl2

Reactivation by cobalt or manganese was further assayed

by incubating the EDTA-inactivated enzyme with different

concentrations of CoCl2or MnCl2for 15–30 min at room

temperature and then measuring the activity for the forward

reaction via the NADH oxidation-based assay as described

above, except that no CoCl2 was added to the reaction

mixture To determine if reactivation is immediate, another

set of reactions was performed, in which each assay was

started by the addition of the EDTA-treated enzyme to

several cuvettes each containing the complete reaction

mixture and the metal at different concentrations

Effect of anions

The assay based on PEP production was the preferred

method for these measurements, because the use of a single

linking enzyme (ENO) resulted in a system that was easier

to interpret, especially since anions such as Cl–and SO42–

are known to strongly affect the kinetics of rabbit muscle

PYK [18] When examining the effect of KCl, this salt was

not added at 50 mM as described above for the standard

assay mixture but at variable concentrations For the effect

of (NH4)2SO4, the auxiliary enzyme enolase [purchased as a suspension in 2.8M(NH4)2SO4] was partially desalted by removing the supernatant after a brief centrifugation prior

to the assay In all cases the assay was started by the addition of the C-LmPGAM

Chemical modification of histidines Diethyl pyrocarbonate (DEPC, Sigma) was diluted 1 : 10

in acetonitrile and stored as 1 mL aliquots at 4C Its concentration was measured by the production of N-carbethoxyimidazole after the reaction of an aliquot with

10 mM imidazole and the consequent change at A240 (3200M )1Æcm)1)

C-LmPGAM was purified and desalted as described above; in a 1 mL reaction, 180 lg of protein (2.9 lM) were allowed to react for 60 min with 0.1 mM DEPC (35-fold molar excess) After different periods of time, 10-lL aliquots were taken from the reaction mixture and diluted 200-fold

in ice-cold 0.1MHepes (pH 7.6), 0.1 mMCoCl2and 25 mM

imidazole to stop the reaction A 0.12 mL sample of each diluted aliquot was used to determine the remaining PGAM activity by the ENO–PYK–LDH coupled assay Two reactions were performed by the addition of DEPC after a

5 min preincubation of the mixture at room temperature with either 0.1 mMEDTA or 0.1 mMCoCl2 Only aceto-nitrile without DEPC was added to the control samples Another set of reactions was performed with 8 lM of purified and desalted protein and 50 lM DEPC (sixfold molar excess) in a total volume of 0.1 mL containing 50 mM

Hepes pH 7.6 and 250 mMNaCl, and PGAM inactivation was followed for 30 min Prior to the addition of DEPC, these mixtures were incubated either with no addition or

in the presence of 0.1 mM EDTA, or 1.5 mM 3PGA, or 0.5 mMCoCl2, or 2 mMMgCl2, or 2 mMMnCl2for 5 min

at room temperature

Identification of modified residues Desalted enzyme was incubated with only acetonitrile or allowed to react with DEPC in the presence or absence of

9 mM3PGA A second set of samples was first inactivated with EDTA which was removed by centrifugation of the protein solution through Sephadex G-50 packed in 1 mL syringes as described [19] The resulting desalted, EDTA-treated enzyme was incubated for 45–60 min at 4C with either no addition or with 25 lMof CoCl2, or 125 lM of MnCl2and then treated with DEPC at room temperature for 12 min Specific activities were checked before and after the filtration, EDTA treatment, CoCl2 reactivation and acetonitrile or DEPC treatment

All samples were subsequently denatured by adding

1 vol 8M urea and incubating for 30 min; the urea concentration was then decreased to 2Mby dilution with 0.2MNH4HCO3and the proteins were digested overnight with 1 lg sequencing grade trypsin at 30C The digestion was stopped by adding trifluoroacetic acid to a final concentration of 0.1% (v/v) The peptides were analysed using fully automated capillary LC-MS/MS Peptides were captured and desalted on a peptide trap (1 mm· 8 mm, Michrom Bioresources) under high flow rate conditions (57 lLÆmin)1) with 1% (v/v) acetonitrile in 0.05% (v/v)

formic acid Separation was performed on a reversed-phase

BioBasic C18 capillary column (0.180 mm· 150 mm,

Thermo Hypersil-Keystone, Runcorn, UK) A linear

10–60% acetonitrile gradient in 0.05% aqueous formic acid

over 100 min was used at a flow rate of 3 lLÆmin)1after

splitting

MS data were acquired using a LCQ Deca XP Plus ion

trap mass spectrometer (ThermoFinnigan) in

data-depend-ent MS/MS mode [20] Dynamic exclusion enabled

acqui-sition of MS/MS spectra of peptides present at low

concentration even when they had coeluted with more

abundant peptides Peptides were identified from the MS/

MS data usingTURBOSEQUEST(ThermoFinnigan) database

search engine or manually with the help of XCALIBUR

software (ThermoFinnigan) Search parameters

incorpor-ated a mass difference of 72.00 atomic mass units for

N-carbethoxyhistidine vs nonmodified histidine

Abun-dance of each peptide species was estimated by their relative

signal intensity and by their peak area after integration

Results

PGAM activity inL mexicana

An initial attempt to measure the mutase activity in

Leishmania lysates was performed with the ENO–PYK–

LDH coupled assay With this method a high background

of NADH oxidation was detected in all fractions, and the

PEP production assay was therefore preferred for locating

the mutase activity Figure 1A shows the activity upon the

addition of 3PGA in the presence of 0.55 U of ENO and a

sample of each cell fraction Under these conditions, the

main reaction monitored should be the PGAM-and

ENO-coupled PEP production These experiments located the

PGAM activity in the cytosol of Leishmania, similar to

previous findings in T brucei [7,21]

In order to attribute the activity to either a

cofactor-dependent or -incofactor-dependent mutase, 5 mL of cytosolic

fraction (high-speed supernatant fraction, S40) were

dia-lyzed A 10 kDa cut-off membrane was used to remove any

potentially interfering metabolites while preserving all

enzymes originally present in the cytosol The specific activity

was not lowered by this deprival of any

2,3-bisphosphogly-cerate that might have been present in the parasite’s cytosol;

on the contrary it was significantly increased, from

780 ± 6 nmolÆmin)1Æmg protein)1 to 1250 ± 75 nmolÆ

min)1Æmg protein)1(Fig 1B) This activity did not increase

when 2,3-bisphosphoglycerate was added to the assay, in

contrast to that of the mammalian d-PGAM that was

enhanced 300% by the addition of its cofactor The increase

of the mutase activity of the Leishmania fraction might be

explained by the presence of 0.1 mMCoCl2in the dialysis

buffer, in line with the fact that i-PGAMs are

metallo-enzymes (see section Requirement for metal ions below)

Further support for the parasite enzyme’s nature as a

metalloprotein is provided by the observation that its activity

is highly sensitive to EDTA, similar to that of purified,

bacterially produced C-LmPGAM (production described

below) The cytosolic mutase activity showed resistance to

Na2VO3, as has been reported previously for i-PGAMs

[22,23], whereas, under similar conditions, the mammalian

enzyme was inhibited by > 90% These data together

confirm the existence of an i-PGAM in Leishmania, present only in the cytosol, and the absence of any detectable d-PGAM activity in this organism

Cloning and sequence ofLmPGAM Two phage k clones of a L mexicana genomic library hybridized with our T brucei i-PGAM probe and yielded identical Southern blot results After shortening the kDNA

by three restriction digestions, the sequencing of the resulting 3.7-kb fragment of L mexicana DNA that was still recognized by the heterologous probe revealed an ORF

of 553 codons with homology to the T brucei enzyme (73.6% identity, Fig 2) The predicted encoded protein possessed a calculated molecular mass of 60 723.38 Da and

an isoelectric point of 5.26 A phylogenetic analysis clustered the new amino acid sequence together with the T brucei i-PGAM and next to the enzymes of vegetal origin, while

Fig 1 PGAM activity in L mexicana (A) PGAM activity in different subcellular fractions of L mexicana promastigotes Activities are expressed as total units in each fraction divided by total protein content

of the lysate S0.5, cell extract (supernatant after removal of silicon carbide); S3.5, cellular extract; P3.5, nuclear fraction; P6.5, large-granular fraction; P11, small-large-granular fraction; P40, microsomal fraction; S40, cytosolic fraction (B) Effect of various treatments (for a detailed description see Experimental procedures) on the PGAM activity in, respectively, the cytosolic (S40) fraction of L mexicana promastigotes, purified bacterially produced C-LmPGAM and com-mercially available rabbit muscle d-PGAM Dotted columns show results before treatment and grey columns, after treatment To assay the effect of EDTA, the mutase was preincubated with 5 m M of this compound and then diluted in the reaction mixture to a final con-centration of 0.25 m M EDTA and 1 m M MgCl 2 in order to avoid EDTA interfering with the (Mg2+-dependent) ENO activity.

having a larger distance to the bacterial ones (not shown

here, but see [7,24] and the URL quoted in the latter

reference)

Bacterial production and purification ofLm PGAM

The LmPGAM gene was fused with a sequence coding for a

short His-tag at the protein’s C terminus using plasmid

pET28 for its expression in E coli Lysates of transformed

E coli BL21 cells showed, upon induction of protein

production, a strong band by SDS/PAGE with the expected

molecular mass of 60 kDa that is not seen in control cells

Approximately half of the protein appeared to be insoluble,

presumably in inclusion bodies, after conditions were

established for its optimal production in soluble form The

soluble cell fraction was taken A single passage through a

metal affinity column resulted in a highly pure (as assessed by

SDS/PAGE; data not shown) and active protein In a typical

expression and purification round, a 50 mL culture yielded

14–20 mg of total soluble protein with a PGAM activity of

22 UÆmg protein)1 This was purified approximately 20-fold

for a final recovery of 0.9–1.2 mg of pure LmPGAM In this

way, approximately 5–9% of all protein found in the soluble

fraction of bacterial lysates corresponded to the Leishmania

enzyme The purified protein had a specific activity of

419 ± 4 UÆmg protein)1for the conversion of 3PGA to 2PGA, as measured by the NADH oxidation method Protein stability

NaCl, imidazole and glycerol were removed from the purified enzyme by gel filtration to determine subsequently the effect of different additives on its stability during storage

at 4C (Fig 3A) In spite of the fact that desalting showed

no effect on LmPGAM activity when measured immedi-ately after the elution, the activity decreased rapidly when

no stabilizer was added The highest stabilizing effect was observed in the presence of NaCl, CoCl2, imidazole or glycerol By comparison of the curves in Fig 3A, we concluded that glycerol, when present together with NaCl, exerted some destabilizing effect Therefore, the preferred storage conditions included only NaCl, CoCl2and imida-zole and the protein retained 80–100% of its original activity after 1 month (data not shown)

Kinetic parameters Kinetic constants were determined using freshly purified and stably stored, bacterially produced protein The meas-urement of NADH oxidation by coupling the reaction to

Fig 2 Multiple alignment of representative i-PGAM sequences Residue numbering is according to the LmPGAM sequence Annotation of secondary structure elements is according to the B stearothermophilus i-PGAM structure (1EJJ.pdb) and is depicted berneath the alignment: cylinders, a-helices; arrows, b-strands Boxes indicate amino acids conserved in all enzymes analysed (these included all the i-PGAMs annotated in SwissProt except for the archaebacterial ones; see text) Bold, amino acids within 5 A˚ of 3-PGA according to 1EJJ.pdb; 7 indicates amino acids within a 7 A˚ radius, where two substitutions are observed: B.s.A461fiL.m.S494 and B.s.E334fiL.m.Q355 Underlining indicates insertion typical

of plant and trypanosomatid i-PGAMs The amino acids involved in chelation of metal ions are indicated with a circle d: 1, corresponding to Mn1 and 2, to Mn2 in 1EJJ.pdb ., serine presumably involved in the phosphoenzyme intermediate Between arrows (above the alignment, at residues Met395, Pro501), metal-chelating motif recognized in the metalloenzyme superfamily (PFAM01676); shadowed, consensus amino acids of this motif according to the Pfam database (including archaebacterial enzymes).

ENO, PYK and LDH was the preferred method for the

characterization of the forward (glycolytic) reaction,

because of the essentially irreversible nature of this assay

(–DG 60 kJÆmol)1); consequently, no or little product

inhibition was observed and the maximal (initial) velocity

was maintained for a long time (2–5 min, with

SD ± 0.001) For both the forward and reverse reaction,

determination of Kmand Vmaxby direct fitting of the data

by the Michaelis–Menten equation gave virtually identical

results to those obtained from Hanes, Lineweaver–Burk or

Eadie–Hofstee plots With regard to the forward reaction,

the enzyme has a K ¼ 0.27 ± 0.02 mMfor 3PGA and a

kcat¼ 434 ± 54 s)1 For the reverse reaction, the Km¼ 0.11 ± 0.03 mMfor 2PGA and the kcat¼ 199 ± 24 s)1 The enzyme showed a similar pH optimum for both directions, located between pH 7.5 and 8.2 (data not shown) The pH–activity profile is broader for the forward reaction with > 75% of maximal activity between pH 6.75 and 8.75 A strong sensitivity of B megaterium i-PGAM to low pH was reported before and shown to be related to its interaction with essential Mn2+ions [25] This was inter-preted as a physiologically important pH-sensing mechan-ism of the enzyme associated with its role in triggering spore formation and germination [25,26] The pH–activity profile

of L mexicana i-PGAM shows effectively a very steep slope in the range between pH 6.0 and 7.4 for the reverse reaction The notably higher tolerance for low pH values observed in the forward reaction might be due to the higher concentration of CoCl2used in this assay In order to avoid the formation of a cobalt precipitate under the reducing conditions of the reverse reaction assay, the concentration

of this metal was kept at only 10 lMwhich is 10 times lower than in the forward one

Effect of anions The effect of salts on LmPGAM activity was determined for the forward reaction (Fig 3B) Only a minor effect of the concentration of salts (ammonium sulphate, KCl) on the activity was observed Solely PO43–was able to inhibit the reaction significantly at relatively low concentrations When, in a single assay in the presence of 100 mMpotassium phosphate, five times more substrate (25 times the Kminstead

of five) was used, the activity was restored to 80% of the maximum velocity (instead of 65%), reinforcing the likeli-hood that PO43–exerts competitive inhibition The relatively low effect of the anions is a major difference compared to what was observed for cofactor-dependent PGAMs, where all ions had a considerable effect For example, the apparent Michaelis constants for the substrates were reported to increase about 10-fold in the presence of 400 mMKCl [27–29] Requirement for metal ions

Figure 4A shows the change of LmPGAM activity when the enzyme is incubated at 4C with 1 mM EDTA for different periods of time Treatment with this metal chelator inactivated the enzyme by > 90% only at pH 8.0, and the presence of the substrate 3PGA at concentrations up to

10 mM showed no significant influence on this loss of activity The fully inactivated samples were diluted 25-fold and incubated with different divalent metal salts Only CoCl2 was able to reactivate the enzyme A similar experiment showed the concentration dependency of this reactivation by cobalt and the inability of manganese to induce the recovery of LmPGAM activity even at higher concentrations (Fig 4B) Notably, 1 mM MgCl2 was pre-sent in each assay, and therefore this metal ion appeared on its own also unable to restore the mutase activity after incubation with EDTA The results shown in Fig 4 were reproduced by similar experiments where the EDTA and EDTA–metal complexes were removed by passage through

a desalting column prior to the reactivation assays both with

Co2+and Mn2+ Also a combination of both metal ions

Fig 3 Biochemical properties of C-LmPGAM (A) Stability: the

activity of the enzyme was assayed after storage at 4 C for different

periods of time in the presence of different agents The protein

con-centration was 0.10 mgÆmL)1in 0.1 M TEA pH 7.6 Additions: m,

none; h, 0.5 M NaCl; s, 0.5 M NaCl and 20% (v/v) glycerol; ·, 0.5 M

NaCl, 20% glycerol and 0.1 m M CoCl 2 ; +, 0.5 M NaCl, 20% glycerol,

0.1 m M CoCl 2 and 25 m M imidazole (B) Effect of anions: m,

(NH 4 ) 2 SO 4 ; ·, KCl; s, potassium phosphate; d, potassium phosphate

plus 6.5 m M 3PGA (instead of 1.5 m M as in the standard assay).

was tested but this did not lead to higher specific activities

than obtained with cobalt ions alone

Chemical modification of histidines

DEPC within the pH range 5.5–7.5 is reasonably specific

for reaction with histidine residues [30] Therefore, the

irreversible carboethoxylation by DEPC has been used for the identification of essential His residues in many different enzymes [31,32] among which is castor plant i-PGAM [33]

It has also been used for the characterization of histidine-containing metal-binding sites [34] As DEPC also hydro-lyses spontaneously in water, some enzyme activity may be retained when such residues are not easily accessible for the compound An initial assay with a 35· molar excess of DEPC over protein and in the presence of 0.1 mMEDTA resulted in 95% irreversible inhibition of the LmPGAM activity, with 75% being lost in the first 5 min of incubation (Fig 5A) In contrast, if no EDTA was added, the

Fig 4 Metal dependency of C-LmPGAM activity (A) Effect of 1 m M

EDTA with time: m, pH 6.2; j, pH 7.0; d, pH 8.0; r pH 8.0, 9.3 m M

3PGA The inset bar diagram shows the relative values of activity

before EDTA treatment (Ctrl), after 6h 15 min at pH 8, 1 m M EDTA

(EDTA), and after 15 min of incubation of the EDTA treated enzyme

in the presence of 100 l M of MnCl 2 (Mn), FeSO 4 (Fe), CoCl 2 (Co),

NiCl 2 (Ni), CuSO 4 (Cu) or ZnCl 2 (Zn) The horizontal line indicates the

background activity without enzyme (B) Effect of MnCl 2 and CoCl 2 :

h, EDTA-treated enzyme after preincubation with MnCl 2 at the

indicated concentrations for 15–30 min at room temperature; r,

reactivation by CoCl 2 either by adding it, at different concentrations,

directly to the assay mixture without preincubation (grey line) or after

preincubating the enzyme with the metal for 15–30 min at the

con-centrations indicated (black line) All assays were performed for the

forward reaction, using the NADH oxidation method All points are

means of replicate experiments For incubation with CoCl 2 , four

dif-ferent experiments were performed at different enzyme concentrations.

Fig 5 Irreversible inhibition by diethyl pyrocarbonate (A) Rates of enzyme inhibition at DEPC : protein molar ratio equal to 35 (fast condition) j, Control with only acetonitrile; d, DEPC alone; s, DEPC plus 0.1 m M EDTA (B) Rate of inhibition at a DEPC : protein molar ratio equal to 6 : 1 (slow condition) d, DEPC alone, a simple exponential curve fits the 5 first min of irreversible inhibition;

s, DEPC plus 0.1 m M EDTA, a double exponential fits best the first

5 min and also predicts the result at 10 min; ·, DEPC plus 1.5 m M

3PGA; +, DEPC plus 1.5 m M PGA and 0.1 m M CoCl 2 The essential residues are protected by 3PGA (whether additional Co2+is present or not); the data corresponding to both incubation with substrate and with substrate plus cobalt were fit together by an equation for a straight line.

inhibitory reaction was halted after 20 min of incubation

and 40% of the original activity remained even after 1 h

In order to compare the protective effects of different

ligands, assays were performed with a smaller excess of

DEPC to slow down the inactivation For examining the

kinetics of the inhibition, only the first 5 min of the reaction

were taken into account, since the DEPC concentration

cannot be considered constant for longer time periods

Irreversible inhibition of the desalted enzyme followed a

simple exponential decay for the initial 5 min of incubation

with DEPC (Fig 5B) An attempt to fit these points by a

double-exponential curve gave an equation with two

virtually identical negative components indicating that a

simple exponential equation describes these results properly

When the enzyme activity was monitored over periods of

10 min or longer, an arrest of the inhibitory reaction was

evident This can be attributed to the rapid decrease of

DEPC concentration, via spontaneous hydrolysis as well as

its reaction with essential and nonessential residues In three

more experiments that were performed in the presence of

either an excess of cobalt, magnesium, or manganese ions,

similar curves were observed with no quantitatively

signi-ficant differences (data not shown) In contrast, chelation of

divalent metal ions by incubation with EDTA made the

enzyme more susceptible to the inhibition by DEPC In this

case, the observed results were best fitted by a

double-exponential decay curve Both equation parameters were

negative, indicating the occurrence of two (groups of)

inhibitory reactions The presence of 3PGA at a

concentra-tion equal to approximately five times the Kmrendered the

enzyme virtually refractory to inactivation by DEPC This

indicates that the residues whose modification led to

inhibition when substrate was absent are most likely

localized in the active site

Identification of modified residues

In order to identify the active-site residues which are

susceptible to chemical modification but protected in the

presence of substrate and metal ions, we complemented the

DEPC experiments with trypsin digestion of the samples,

followed by analysis of the peptides by LC-MS/MS The

average protein coverage was 60% and 14 histidine residues

out of 18 present in the enzyme could unambiguously be

identified by MS/MS fragmentation of their corresponding

peptides First, a control sample (acetonitrile) was analysed

in order to identify the His-containing peptides In a second

experiment, a DEPC-treated sample was analysed and the

corresponding peptides with DEPC-dependent modification

were identified taking into account a mass increase of

72.00 Da per modified residue A total of 10 His residues

were found to be modified (Table 1), although none of them

was stoichiometrically labelled as the corresponding

unmodified peptides were still present It should be noted

that the enzyme was fully active just before DEPC treatment

and remained stable in the presence of only acetonitrile

during the time of incubation (12 min, room temperature),

thus indicating that the observed irreversible inhibition was

entirely caused by the reaction with DEPC In addition, two

samples were treated in the presence of a substrate (3PGA)

concentration which in earlier experiments, where PGAM

activity was assayed, showed significant protection

(100 ± 16% and 82 ± 3% of original activity) The results are summarized in Table 1, where all histidines present in native LmPGAM are listed It is indicated in the table which

of these residues were modified or protected in the experiments with DEPC and substrate + DEPC The location of the residues with respect to the active site or the surface was identified by sequence alignment with the

B stearothermophilusenzyme, of which the crystal structure

is known [35,36] and by examining a recently solved, unpublished structure of LmPGAM (B Poonperm, M Walkinshaw and L A Fothergill-Gilmore, unpublished data) Figure 6 shows the spatial distribution of all conserved histidines, together with two important active-site residues, Lys357 and Ser75 All surface histidines were modified with the sole exception of His37 In the active site, two histidines, His136 and His467, were modified by DEPC but protected from this reaction by the presence of substrate, while two others, His360 and His429, were not accessible under any condition Interestingly, His160 was apparently protected by the binding of 3PGA in spite of being located far away from the active site In the inhibited sample, two modifications were found in the peptide comprising both His60 and His79, whereas with substrate present only indications for modification of a single His were obtained However, it was not possible to distinguish which of these His residues was protected in the latter case

Table 1 Modification of LmPGAM residues by DEPCand protection

by the substrate 3PGA Histidines located closer than 10 A˚ from the substrate are considered as part of the active site and those with accessibilities higher than 10% as belonging to the protein surface Results of site-directed mutagenesis in castor plant i-PGAM [33] are noted aside; percentages indicate the remaining activity after HisfiAla mutations Histidines 53, 231 and 233 are not included since their corresponding tryptic peptides were too small to be seen and/or retained by the C 18 column.

Residue Inhibiteda Protecteda Active site HfiA

His429 (M1) – – Yes Insoluble His496 (M1) n.d n.d Yes 0%

His467 (M2) + (–) Yes Insoluble His9 + + No, surface n.a.

His37 – – No, surface 100% His47 + n.d No, surface Not conserved His60 + or 79 No, surface 72%

His114 + n.d No, surface Not conserved

His148 + + No, surface Not conserved

His377 + n.d No, surface Not conserved

a Positive signs indicate that the modified peptide was present and its sequence confirmed by LC-MS/MS; negative signs indicate that only the unmodified peptide was detected Parentheses indicate that His467 was detected as modified but in a very low amount n.d., A peptide of the corresponding mass was not detected or, if detected, its sequence was not confirmed by MS/MS analysis n.a., Nonassayed mutations; insoluble, cases where the mutated protein was insoluble.

Another set of experiments was performed in which

LmPGAM was incubated with EDTA followed by

desalt-ing through Sephadex G-50 columns and subsequent

addition of Co2+and Mn2+salts in order to determine

the influence of the presence of metal ions on the

accessi-bility of the active-site histidines Table 2 shows the results

for the different EDTA-treated enzyme samples Clearly,

the histidines corresponding to the first metal ion-binding

site (namely His429 and His496) were not modified under

any condition His467 corresponds to the second metal

ion-binding site and it was modified to the same extent in

both samples either with no addition or with MnCl2

In contrast, the sample that was partially reactivated

(54 ± 2% of original activity) by incubation with CoCl2

showed a significant protection of His467, evidenced by a

significantly lower chromatographic peak for the

corres-ponding peptide mass

Discussion

As shown previously for T brucei [7], L mexicana also

contains an i-PGAM gene Furthermore, aBLASTsearch in

the genome database of L major strain Friedlin (http://

www.geneDB.org/) identified on chromosome 36 an ORF

encoding an amino acid sequence with 92% identity with

that of the LmPGAM reported in this paper A similar

search using the yeast d-PGAM sequence did not yield a

significant match in the trypanosomatid databases

The PGAM activity in cultured L mexicana promasti-gotes is essentially localized in the cytosolic fraction and corresponds exclusively to a cofactor-independent enzyme,

as shown by the effects of cobalt, 2,3-bisphosphoglycerate, EDTA and vanadate We have developed a bacterial expression system and purification protocol for a LmPGAM with an eight-residue long C-terminal tag (containing six His residues) that resulted in a yield of

 1 mg of pure protein per 50 mL of culture Conditions for stable storage and optimal activity assays of the enzyme were established The experimentally determined data were excellently fitted by Michaelis–Menten equations, allowing accurate calculation of the kinetic constants

Stability assays showed that without an excess of CoCl2

in solution, the activity of LmPGAM decreased within days

It is important to note that passage of the enzyme through a desalting column equilibrated with only TEA buffer did not affect its specific activity when measured immediately afterwards in an assay buffer containing CoCl2 (see Experimental procedures) However, overnight incubation

at 4C (or few hours at room temperature) in the presence

of EDTA led to irreversible inactivation of a major proportion of the enzyme preparation (not shown) These observations indicate that the enzyme contains at least one essential metal ion that is in equilibrium between its protein-bound and solute form, and that the metal-deprived enzyme slowly denatures The Co2+ions are, most likely, involved

in the catalytic activity (see below) but their presence seems also important for the correct conformation of the LmPGAM active site and consequently the stabilization

of the enzyme’s overall structure

The stabilizing effect of imidazole, being synergistic with the effect of CoCl2, underlines the importance of a soluble cobalt reservoir Imidazole as a metal ligand favours the desirable 2+valency and hampers the irreversible formation

of Co(OH)2 precipitates, always observable as pink dust after a few days even at concentrations as low as 100 lM

when no imidazole was added It has been reported

Fig 6 Spatial distribution of the conserved histidines in LmPGAM.

The LmPGAM sequence was threaded in the B stearothermophilus

structure (PDB code EQJ) as described in Experimental procedures.

The substrate (product) 2PGA is displayed in thin balls-and-sticks

format, while amino acids are depicted with thick sticks Lys357 and

Ser75 were also included in the picture because of their relevance to our

study Spheres M1 and M2 correspond, respectively, to Mn1 and Mn2

in the EJJ structure By analogy with the B stearothermophilus

structure, M1 is proposed to be coordinated by His429 and His496,

while His467 interacts with M2 His160 is located approximately 25 A˚

from M1 and 20 A˚ from the bound 2PGA.

Table 2 Activity of LmPGAM and modification of His residues in-volved in metal binding after EDTA treatment and subsequent incubation with Co 2+ or Mn 2+ and DEPCtreatment EDTA, remaining activities correspond to activities after incubation with EDTA, passage through Sephadex G-50 and, when indicated, incubation with cobalt or man-ganese chloride; DEPC, remaining activity was measured for the co-balt reactivated sample Brackets indicate that His467 was found modified in a significantly lower amount ND, Not determined.

EDTA

EDTA + CoCl 2

EDTA + MnCl 2

Activity (%) EDTA – Remaining activity

10 ± 0% 54 ± 2% 7 ± 2% DEPC – Remaining

activity

DEPC Modification