Folylpolyglutamate Synthase from Plasmodium Falciparum; a Potential Novel Target for Antimalarial Antifolate Inhibition

Molecular & Biochemical Parasitology 172 (2010) 41–51

Contents lists available at ScienceDirect

Molecular & Biochemical Parasitology

Characterisation of the bifunctional dihydrofolate synthase–folylpolyglutamate synthase from Plasmodium falciparum; a potential novel target for antimalarial antifolate inhibition

Ping Wang a, Qi Wang a, Yonghong Yang b,c,1, James K. Coward b,c, Alexis Nzila d, Paul F.G. Sims a, John E. Hyde a,∗

a Manchester Interdisciplinary Biocentre, Faculty of Life Sciences, University of Manchester, 131 Princess Street, Manchester M1 7DN, UK b Department of Medicinal Chemistry, University of Michigan, 930 N. University, Ann Arbor, MI 48109-1055, USA c Department of Chemistry, University of Michigan, 930 N. University, Ann Arbor, MI 48109-1055, USA d KEMRI, Wellcome Trust Collaborative Research Programme, Kiliﬁ 80108, Kenya

article info abstract

Article history: Unusually for a eukaryote, the malaria parasite Plasmodium falciparum expresses dihydrofolate synthase Received 29 January 2010 (DHFS) and folylpolyglutamate synthase (FPGS) as a single bifunctional protein. The two activities con- Received in revised form 15 March 2010 tribute to the essential pathway of folate biosynthesis and modification. The DHFS activity of recombinant Accepted 16 March 2010 PfDHFS–FPGS exhibited non-standard kinetics at high co-substrate (glutamate and ATP) concentrations, Available online 27 March 2010 being partially inhibited by increasing concentrations of its principal substrate, dihydropteroate (DHP). Binding of DHP to the catalytic and inhibitory sites exhibited dissociation constants of 0.50 ␮M and Keywords: 1.25 ␮M, respectively. DHFS activity measured under lower co-substrate concentrations, where data fit- Antifolate inhibitor studies ␮ ␮ Biphasic kinetics ted the Michaelis–Menten equation, yielded apparent Km values of 0.88 M for DHP, 22.8 M for ATP ␮ Folate metabolism and 5.97 M for glutamate. Of the substrates tested in FPGS assays, only tetrahydrofolate (THF) was Methotrexate efficiently converted to polyglutamylated forms, exhibiting standard kinetics with an apparent Km of Phosphinate analogues of folate 0.96 ␮M; dihydrofolate, folate and the folate analogue methotrexate (MTX) were negligibly processed, Substrate specificity emphasising the importance of the oxidation state of the pterin moiety. Moreover, MTX inhibited neither DHFS nor FPGS, even at high concentrations. Conversely, two phosphinate analogues of 7,8-dihydrofolate that mimic tetrahedral intermediates formed during DHFS- and FPGS-catalysed glutamylation were pow-

erfully inhibitory. The Ki value of an aryl phosphinate analogue against DHFS was 0.14 ␮M and for an alkyl phosphinate against FPGS 0.091 ␮M, with each inhibitor showing a high degree of speciﬁcity. This, com- bined with the absence of DHFS activity in humans, suggests PfDHFS–FPGS might represent a potential new drug target in the previously validated folate pathway of P. falciparum. © 2010 Elsevier B.V. All rights reserved.

1. Introduction a million lives each year and causes considerable economic loss to developing countries [1,2]. The antifolates that have been deployed The antifolates are clinically important drugs used to target against malaria target only two enzymes in the folate biosynthe- both bacterial and eukaryotic pathogens, including the apicom- sis pathway of the parasite. Dihydropteroate synthetase (DHPS; plexan, Plasmodium falciparum (Pf), a parasite that still claims over EC 2.5.1.15), which catalyses the synthesis of 7,8-dihydropteroate (DHP) from 6-hydroxymethyl-7,8-dihydropterin pyrophosphate and p-aminobenzoate (pAB), is inhibited by sulfonamide and sul- Abbreviations: DHF, dihydrofolate; DHFR, dihydrofolate reductase; DHFS, dihy- fone drugs, while dihydrofolate reductase (DHFR; EC 1.5.1.3), which drofolate synthase; DHP, dihydropteroate; DHPS, dihydropteroate synthase; DTT, reduces 7,8-dihydrofolate (DHF) to the active cofactor 5,6,7,8- dithiothreitol; FPGS, folylpolyglutamate synthase; GTPCH, GTP cyclohydrolase tetrahydrofolate, is targeted principally by pyrimethamine (PYR) I; HPPK, hydroxymethyldihydropterin pyrophosphokinase; MTX, methotrexate; and cycloguanil. Inhibition of these activities, most commonly in a pAB, para-aminobenzoate; PTPS, pyruvoyltetrahydropterin synthase III; PYR, pyrimethamine; SDX, sulfadoxine; THF, tetrahydrofolate. synergistic combination of sulfadoxine and PYR, disrupts the con- ∗ Corresponding author at: University of Manchester, Faculty of Life Sciences, stant supply of tetrahydrofolate (THF) cofactors required for key Manchester Interdisciplinary Biocentre, 131 Princess St, Manchester M1 7DN, UK. 1-carbon transfer reactions, including synthesis of thymidylate, Tel.: +44 161 306 4185; fax: +44 161 306 5201. which is essential for DNA replication. E-mail address: [email protected] (J.E. Hyde). The long-established efﬁcacy of folate metabolism as a clini- 1 Current address: Carestream Health Inc., 4 Research Dr., Woodbridge, CT 06525, USA. cal target and the widespread resistance to current antifolates and

Fig. 1. P. falciparum DHFS–FPGS: roles, expression of recombinant protein and product analysis. (a) Position (grey boxes) and roles of DHPS and FPGS activities in the folate pathway of P. falciparum leading to 5,6,7,8-tetrahydrofolate (THF). Polyglutamation of folates is thought to occur on their tetrahydro-forms. The addition of one or more glutamate residues by FPGS to THF or its modified forms, produced de novo or from salvaged host folates, is indicated by nGlu. The dotted arrow indicates sequential steps involving the four enzymes shown. (b) SDS-PAGE analysis of purified PfDHFS–FPGS after Ni-agarose affinity and ion-exchange chromatography; (c) mass spectral analysis of the reaction mix after a standard 1 h incubation of the DHFS assay. other classes of antimalarials highlights the importance of iden- major anti-cancer drug and folate analogue methotrexate (MTX) tifying other targets within the P. falciparum folate/thymidylate [17,18]. biosynthesis pathways, which comprise a further seven enzyme Some bacteria, such as Corynebacterium species [19], Neisseria activities in addition to DHPS and DHFR [3]. One activity of gonorrhoeae [20], Mycobacterium tuberculosis [21] and E. coli [22] folate biosynthesis yet to be characterised in malaria parasites express bifunctional DHFS–FPGS molecules that perform both the is dihydrofolate synthase (DHFS; EC 6.3.2.12), which adds an l- first and subsequent additions of glutamate. In E. coli and M. tuber- glutamate residue to the pAB component of DHP, the product culosis, recent work has shown that the active centres of the DHFS of DHPS, to form DHF, the substrate of DHFR (Fig. 1a). DHFS and FPGS functions co-localise closely, at least for addition of the represents a target unique to the parasite, as the human host first glutamate to THF by the latter, but are not identical [21,23,24]. is unable to synthesise folates and lacks this enzyme. Closely However, in L. casei the FPGS has no accompanying DHFS activity related to the activity of DHFS is that of folylpolyglutamate syn- [25] and folate must be salvaged. This is also the case in mammals, thase (FPGS; EC 6.3.2.17), which adds further glutamate residues including humans [26,27], where pre-formed folate is an essential to reduced folate monoglutamates by ␥-linkage (Fig. 1a), with nutrient. In eukaryotes that can synthesise folate de novo, such as the number of residues incorporated varying among organisms fungi and plants, both activities are present but are usually encoded [4,5], ranging from an average of 3 in Escherichia coli [6] up to by separate genes [28,29]. However, we have demonstrated previ- as many as 9 in Lactobacillus casei [7] and mammals [6].InP. ously that a single protein from P. falciparum carries both DHFS and falciparum, the pentaglutamate form of 5-methyltetrahydrofolate FPGS activities, the first example of a bifunctional enzyme of this has been identified as the predominant polyglutamylated folate type from a eukaryotic organism [30]. metabolite [8,9], demonstrating the importance of FPGS catal- The critical dual role of parasite DHFS–FPGS in both the biosyn- ysis in these organisms as well. Such conjugated forms are thesis and modification of folates, and the absence of DHFS activity apparently ubiquitous and the crucial role of polyglutamylation in humans suggest the possibility that parasite-specific inhibitors has been demonstrated by studying organisms mutated in their targeted to this molecule might be feasible and effective. We there- fpgs genes. For example, CHO cells mutant in this gene require fore undertook a detailed study of PfDHFS–FPGS with respect to supplementation by the end-products of folate metabolism and its kinetic properties, substrate specificities and susceptibility to exhibit much reduced levels of intracellular folates, predominantly the antifolate drug MTX, as well as to novel inhibitors based on as monoglutamates [10,11]. Similarly, the MET7 gene encod- phosphinic acid analogues of folic acid. ing FPGS in Saccharomyces cerevisiae is essential for methionine biosynthesis and the maintenance of mitochondrial DNA [12].In 2. Materials and methods mammals and plants several folate-dependent enzymes exhibit much higher affinity for polyglutamylated folates compared to 2.1. Reagents their monoglutamylated equivalents [4,13,14]. Another function of polyglutamylation is to prevent folates from leaking through Specialty reagents were obtained commercially as fol- the cell membranes and sub-cellular compartments by substan- lows: l-[U-14C] glutamic acid (238 mCi/mmol), 3H-folinic acid tially increasing the negative charge they carry [10,15,16] and in (30 Ci/mmol) and 3H-methotrexate (31.8 Ci/mmol) (Moravek human cells, polyglutamylation by FPGS has been shown to have Biochemicals, Inc., California); DE81 anion-exchange chromatog- a critical role in the cellular retention and enzyme targeting of the raphy paper (Whatman International Ltd., UK); DHF, THF, folinic P. Wang et al. / Molecular & Biochemical Parasitology 172 (2010) 41–51 43 acid and DHP (Schircks Laboratories, Jona, Switzerland); 2- 2.4. Metabolic labelling of parasite cultures with 3H-folinic acid mercaptoethanol, sodium hydrosulfite (dithionite), folic acid, and 3H-methotrexate ATP, BSA, l-glutamic acid, and dithiothreitol (DTT) (Sigma), Overnight ExpressTM Instant TB Medium (Merck), Ni-NTA resin 0.5 ␮Ci/ml of either 3H-folinic acid or 3H-methotrexate at the (Qiagen Ltd., UK). The E. coli expression host used was BL21(DE3) same specific activity was used to label parallel 50 ml synchronous (Novagen). The isolates of P. falciparum used were K1, FCB, V1/s, cultures of P. falciparum FCB strain. Label was incubated with Fcr3, as well as the cloned line 3D7. The aryl phosphinate ring-stage (6–10 h) parasites, which were harvested at the late folate analogue, 2-[[[4-[N-[(2-amino-3,4-dihydro-4-oxo-6- trophozoite stage 20 h later. The labelled products were extracted pteridinyl)methyl]amino]phenyl](hydroxy)phosphinoyl]methyl] by the boiling method and analysed using HPLC on a C18 column pentane-1,5-dioic acid (compound 1) [31] and the alkyl phosphi- eluted with an acetonitrile gradient as described [38]. nate folate analogue, 2-[[[3-[[4-[[(2-amino-3,4-dihydro-4-oxo- 6-pteridinyl)methyl]amino]benzoyl]amino]-3-carboxypropyl]hy- 2.5. MALDI-ToF mass spectrometry of DHFS–FPGS reaction droxyphosphinyl]methyl]pentane-1,5-dioic acid (compound 2) products [32,33] were synthesised in the laboratory of JKC. The matrix used was ␣-cyano-4-hydroxycinnamic acid, which 2.2. Expression and protein purification was dissolved in acetonitrile:water (1:1, v/v), 0.1% formic acid (v/v) at a concentration of 10 mg/ml. The reaction mix was partially A verified cDNA fragment encoding the entire dhfs-fpgs gene purified with a C18 SPE device. One ␮l of the eluted material in [30,34] was cloned from the K1 isolate of P. falciparum into pET22b 50% acetonitrile, 0.1% formic acid was mixed with the same vol- (Novagen, UK) using the NdeI and BamHI sites and the con- ume of matrix solution and 1 ␮l of the mix placed onto the target struct transformed into E. coli BL21(DE3) host cells. Production of and allowed to air dry. Spectra were acquired with an AXIMA-CFR DHFS–FPGS was carried out by an autoinduction procedure [35] in MALDI-ToF spectrometer (Kratos Analytical Ltd., Manchester, UK), Overnight Express instant medium (Novagen, UK). Incubation was in positive reflectron mode with a power setting of 180. The spec- at 37 ◦C overnight and then at 18 ◦C for a further 24 h. The cell pellet tra were obtained by overlapping 100 profiles. Mass to charge (m/z) was resuspended in 50 mM sodium phosphate buffer, pH 8, with calibration was performed externally using the matrix peaks and 300 mM NaCl at a concentration of up to 200 g wet weight/l, and folyldiglutamate as standards. processed with a high pressure cell breaker (Constant Cell Disrup- tion System, Constant Systems Ltd., UK) at a pressure of 25,000 psi. 2.6. Reduction of pteroyl phosphinates Cell debris was removed by centrifugation and the supernatant collected. The His-tagged DHFS–FPGS was then purified on a Ni-NTA Before evaluation as possible inhibitors, the pterin rings of com- column and bound protein eluted with 250 mM imidazole. The pounds 1 and 2 (Section 2.1) were reduced with mild sodium protein was further purified on a Q-Sepharose high performance hydrosulfite (dithionite) treatment to provide the corresponding ion-exchange column, using a loading buffer of 25 mM Tris–HCl, 7,8-dihydro derivatives [39,40], as confirmed by MALDI-ToF spec- pH 7.5 and eluting with a gradient from 0 to 1 M NaCl in the same trometry of the parent and reduced compounds (negative ion mode, buffer. The activity was monitored by the DHFS assay described − − − 1 as [M−H] = 604.2, H2-1 as [M−H] = 606.2; 2 as [M−H] = 474.4, below and the active peak area collected. Yields were ∼2 mg/l of − H2-2 as [M−H] = 476.4; negligible signals for H4X forms). For this starting E. coli culture. Glycerol (50%, v/v) was added to the protein, reduction, the parent compounds were first dissolved in DMSO at − ◦ which was stored at concentrations of 0.5–1 mg/ml at 80 C for 5 mM then reduced in a 50% aqueous solution containing sodium future use. hydrosulfite (20 mg/ml) and 2-mercaptoethanol (10 mM). Mixes were kept at room temperature for 80 min, and then the com- 2.3. DHFS and FPGS assays pounds aliquoted, analysed and either used immediately or moved to −20 ◦C for short-term storage. A fresh aliquot was used for each A standard reaction mix based on those established for bacterial assay. and eukaryotic DHFS/FPGS assays was adopted [28,36,37] and con- tained 10 mM MgCl2, 5 mM DTT, 200 mM KCl, 100 mM Tris, 50 mM 3. Results glycine, 1 mg/ml BSA, 45 ␮g/ml DHFS–FPGS protein (0.72 ␮M), pH 10. For the three substrates of the reaction, up to 5 mM ATP, up 3.1. Expression of dihydrofolate synthase and folylpolyglutamate 14 to 1 mM l-[U- C] glutamate (diluted from the commercial stock synthase with unlabelled l-glutamate to the required specific activity) and up to 100 ␮M DHP (for DHPS) or THF (for FPGS) were used, depend- The dhfs-fpgs open reading frame obtained as cDNA from P. ing upon the purpose of the assay, as indicated in the text. The falciparum isolate K1 [34] was cloned, expressed and purified as ◦ enzyme reactions were carried out at 37 C for 60 min in triplicate described in Section 2. The purified His-tagged protein ran on SDS- and stopped by transferring reaction tubes into an ice-cold water PAGE close to its predicted molecular mass of 62.4 kDa (Fig. 1b). bath. Reaction mixes were then spotted onto DE81 ion-exchange Identity and purity were confirmed by LC–MS–MS analysis of paper (pre-treated with 0.5 M EDTA, pH 8.0 and then dried), air tryptic peptides from the purified protein (Mascot score = 468, dried and the unreacted label washed away with 125 ml of 40 mM against complete UNIPROT database; 31% sequence coverage, with NaCl, 10 mM Tris–HCl, pH 8.0, four times, 20 min each wash. The fil- no detectable peptides from any other protein, including E. coli ter was then exposed to a storage phosphor screen overnight and DHFS–FPGS). Gel filtration experiments in non-denaturing con- the screen scanned with a quantitative imager (Typhoon Trio Vari- ditions gave an apparent molecular mass of ∼90 kDa (data not able Mode Imager, GE Healthcare, UK). The measured counts were shown), which although significantly higher than the calculated converted to concentrations by comparison with a standard curve mass of a single chain, was too small to represent a dimer, indicat- derived from known amounts of the radiolabelled l-glutamate sub- ing that the plasmodial protein is most likely to be monomeric, in sequently used as substrate in the assays. The enzyme activity data common with all other DHFS/FPGS enzymes thus far reported from was analysed using the Matlab software package. Regression curves bacterial [22,25], fungal [28], plant [41] and mammalian [36,42] were fitted using the least squares routine. sources. 44 P. Wang et al. / Molecular & Biochemical Parasitology 172 (2010) 41–51

Fig. 2. Dependence of PfDHFS activity on DHP, ATP and glutamate with DHP taken as the variable substrate. (a) ATP is kept at a constant high level of 0.5 mM in all the assays in this panel, with glutamate concentration (Glu) set at three different levels ≥75 ␮M. (b and c) As for (a), except ATP is kept at a constant low level of 25 ␮M in (b) and glutamate at a constant low level of 25 ␮M in (c), with stepped increments of glutamate and ATP, respectively.

PfDHFS–FPGS represents a complex system where the closely pendent variable. Thus, this phenomenon is restricted to increases related DHFS and FPGS activities each catalyse reactions in which in DHP concentration only. Control experiments established that there are three substrates, the pterin/folate moiety (DHP in the case the observed kinetics were independent of the order of addition of of DHFS, THF mono- or polyglutamylated forms in the case of FPGS), enzyme and substrates to the reaction mix prior to the incubation ATP and glutamate. To establish initial conditions, the enzyme was period. tested separately for DHFS and FPGS activities using saturating In order to understand the mechanism of the DHFS reaction, the concentrations of the natural substrates DHP and THF, respec- activity data sets were fitted globally with a series of non-linear tively, together with ATP and glutamate. From progress curves regression models, starting from the assumption that the sequence with respect to both enzyme concentration and time, a standard in which the substrates bind is random, rather than ordered, to reaction for both activities containing 0.72 ␮M protein incubated form a quaternary enzyme–substrate complex (Supplementary Fig. for 60 min was adopted. This permitted accumulation of maxi- 1), an assumption that is confirmed below. Initially, substrate inhi- mal counts for accurate quantitation while staying well within the bition with a second DHP molecule competitively binding to the period of linearity (at least 90 min; data not shown). When DHP enzyme was considered. However, though the modified formula was used as substrate, no products with mass greater than that (Supplementary Fig. 2, equation e) did follow the data trend, unlike of DHF (which, because of its lability, was seen primarily as its the Michaelis–Menten equation (Supplementary Fig. 2, equation f), oxidised product folic acid in the MALDI-ToF mass spectrometer, the fit was only improved slightly, mainly at very low DHP concen- theoretical MH+ = 442.4) were detectable (Fig. 1c), demonstrating trations (compare Supplementary Fig. 3e and f). The experimental that under these reaction conditions the DHF produced by the data showed that, although the rate of the reaction decreased DHFS activity is not subject to further glutamylation by the FPGS with increasing DHP concentration, the inhibition was not com- activity. plete, even at high levels of DHP (≥100 ␮M). Therefore, a partial inhibition term, V2 (Supplementary Fig. 2, equation d), was con- 3.2. Kinetics of the DHFS reaction; DHP substrate inhibition and sidered more appropriate. This assumes that a second binding site dependence on co-substrate concentrations exists on the enzyme complex. Binding of a second molecule of DHP would change either the affinity of the enzyme complex to Standard kinetic experiments in which two of the three sub- the normally bound substrate or the Vmax of the enzyme. Intro- strates were present in excess, while the concentration of the third duction of this term considerably improved the fit, especially at was varied, revealed that the DHFS reaction did not follow classi- higher DHP concentrations (Supplementary Fig. 3d). However, all cal Michaelis–Menten kinetics. As seen in Fig. 2a, velocity versus of these curves, which were derived from the above equations [S] curves where ATP was fixed at a high concentration (0.5 mM) without considering the second and third substrates (ATP and and glutamate was also raised to ∼1 mM, displayed character- glutamate), gave at best only imperfect fits to the data sets, sug- istic biphasic kinetics, with fast initial rates that diminished as gesting that the binding of the different substrates may also exert the concentration of DHP was increased. The activity peaked at allosteric effects on the enzyme. Of the equations that took this about 1–2 ␮M DHP. Similar-shaped curves indicative of substrate into account (Supplementary Fig. 2, equations a, b and c, graphed in inhibition were obtained when glutamate was fixed at concen- Supplementary Fig. 3), partial substrate inhibition in the presence trations >50 ␮M and ATP was varied, along with DHP (data not of all three substrates, as in equation a, gave an extremely good shown). However, neither of these co-substrates showed similar fit with an R2 value of 0.95 (Supplementary Fig. 3a). The next best inhibition curves at higher concentrations when taken as the inde- fit (equation b), where fully competitive substrate inhibition in the P. Wang et al. / Molecular & Biochemical Parasitology 172 (2010) 41–51 45

Fig. 3. Comparison of DHFS and FPGS activities under identical conditions of high co-substrate concentrations. ATP is at 5 mM, glutamate at 275 ␮M. The data are ﬁtted to the three-substrate, substrate partial inhibition model of Supplementary Fig. 2, equation a.

presence of three substrates was modelled, gave a much poorer 3.4. Kinetics of the FPGS reaction and competition between the R2 value of ∼0.6 (Supplementary Fig. 3b). Over a broad range of substrates of DHFS and FPGS conditions where concentrations of ATP from 6 ␮M to 5 mM and of glutamate from 25 ␮M to 1 mM were tested in various combina- Similar experiments to the above were conducted using THF as tions, the best-fit model, a, gave a dissociation constant, KA, for DHP substrate to assay the FPGS activity of PfDHFS–FPGS. Unlike the case of 0.50 ± 0.06 ␮M for the substrate binding, and a dissociation con- for the DHFS activity, no unusual kinetics were observed in parallel stant, KABCA, of 1.25 ± 0.37 ␮M for the substrate (DHP) inhibition assays on the same enzyme preparation, either under low or high (n = 15 with R2 > 0.90 in each case). Due to the substrate inhibition substrate concentrations (Fig. 3). Velocity versus [S] curves fitted feature of the system, there is no conventional substrate Km term well to the Michaelis–Menten equation, yielding an apparent Km in this equation. value of 0.96 ± 0.12 ␮M for THF, very similar to the KA and appar- That the extent of the DHP substrate inhibition phenomenon ent Km values for DHP in the DHFS reactions above. This lack of seen above is dependent on ATP and glutamate concentrations is substrate inhibition of the FPGS activity of PfDHFS–FPGS is in con- also clearly seen in Fig. 2b and c. When either ATP or glutamate contrast to the marked substrate inhibition noted previously in studies centrations were lowered to ≤25 ␮M, as in (b) and (c), respectively, with human FPGS [43,44]. such inhibition by DHP was no longer significant. Data obtained Given the close similarity in the reactions catalysed by DHFS under these conditions were found to fit well to the conventional and FPGS, the similarity in the affinities of their substrates and data Michaelis–Menten equation, enabling the calculation of apparent from studies of bacterial DHFS–FPGS proteins suggesting that the Km values for the three substrates, because here the reaction was catalytic sites of these activities are at the very least closely linked found to be first order with respect to DHP concentration. Using [21,24], we measured the rate of glutamylation in the presence of a non-linear fit to the classical Michaelis–Menten equation, we equimolar amounts of both DHP and THF substrates at two concen- determined an apparent Km for DHP under these conditions to be trations differing by an order of magnitude (Fig. 4a and b). In both 0.88 ± 0.26 ␮M, quite closely comparable to the dissociation con- cases, no significant additive effect was seen in the rate of gluta- stants calculated above for DHP binding at higher ATP/glutamate mate incorporation measured relative to those seen for the same concentrations in the partial substrate inhibition model. The appar- concentration of each substrate alone, consistent with there being ent Km values for ATP and glutamate were 22.8 ± 2.7 ␮M and only a single site of glutamylation available for either substrate at 5.97 ± 0.50 ␮M, respectively. any given time.

3.3. Kinetics of the DHFS reaction; substrate binding to DHFS 3.5. Substrate speciﬁcity of the DHFS–FPGS activities

To study the binding of the three substrates to DHFS, this To establish the specificity of PfDHFS–FPGS, equal concentra- activity was further investigated in a series of assays where tions of substrates varying in the oxidation level of the pterin each of the three substrates in turn was taken as the variable moiety were compared, i.e. DHP, DHF, THF and folic acid. The activi- substrate (x-axis), maintaining the concentration of one of the ties were studied under both low and high concentration conditions other two constant and increasing the third in steps to yield of the test substrates and of the co-substrates ATP and glutamate double-reciprocal plots that are indicative of the enzyme mecha- (Fig. 5a and b). Under both of these conditions, the highest level nism. These experiments were done under the above-mentioned of glutamylation activity was seen on DHP, the natural substrate conditions of low substrate concentrations, where the reaction of DHFS, followed by THF, the natural substrate of FPGS. In con- data fit to the Michaelis–Menten equation. All combinations trast, both folic acid and DHF were very poor substrates for the of conditions gave rise to convergent plots rather than paral- polyglutamylation activity of the latter. Especially under low sub- lel lines, as exemplified in Supplementary Fig. 4. These data strate conditions, incorporation of glutamate onto folic acid was confirm that the binding of the substrates DHP, ATP and glu- insignificant. Similarly, although DHF, the product of the DHFS tamate occurs sequentially as depicted in Supplementary Fig. activity, is in the active centre after the DHFS reaction, it seem- 1, giving rise to the quaternary complex anticipated in the ingly cannot be efficiently polyglutamylated by the FPGS activity. modelling exercise, and exclude the possibility of a ping-pong Thus, using the same concentrations of DHF and THF as substrates mechanism. for FPGS, the enzyme produced only 21% under low substrate 46 P. Wang et al. / Molecular & Biochemical Parasitology 172 (2010) 41–51

Fig. 4. Effect of having both DHFS and FPGS substrates together in the same reaction compared to when present alone. From left to right in (a) 10 ␮M DHP, 10 ␮M THF, 10 ␮M DHP+10␮M THF; (b) 100 ␮M DHP, 100 ␮M THF, 100 ␮M DHP + 100 ␮M THF. Glutamylation activity is represented as a percentage of that seen with DHP as substrate in each case. In all cases 5 mM ATP and 275 ␮M glutamate were present. concentrations (25 ␮M of either) and 12% under high substrate con- in the case of folates themselves, are more efficiently retained centrations (100 ␮M of either) of polyglutamylated forms of folate in the cell and also show markedly increased affinity to certain from DHF compared with that produced from THF. Although some folate-dependent enzymes [46], such as thymidylate synthase [47], incorporation onto DHF does occur in the above experiments, the 5-amino-4-imidazolecarboxamide ribotide transformylase (AICAR concentrations used are artificially high in both cases—about two formyltransferase), involved in purine biosynthesis [48],orthe orders of magnitude higher than the concentration of DHF that enzymes of methionine biosynthesis [12]. The relevance to malaria could be produced in our assays from the same concentration of is that the use of MTX has been advocated as a cheap antimalar- DHP via DHFS over a 1 h incubation period. This is consistent with ial that could be safely used in doses far lower than necessary the MALDI-ToF mass spectrometry analysis of the reaction products in cancer chemotherapy, and it has thus been envisaged that the (Fig. 1c), where no trace of polyglutamylated products was seen parasite could also convert MTX into polyglutamylated forms and when DHP was used as substrate under these conditions. These therefore enhance its efficacy by broadening its range of targets, experiments confirm that the reduction status of the folate moiety as in the case of cancer cells [49]. We therefore tested MTX in two is critical for the polyglutamylation reaction, with the fully reduced ways. First, using radiolabelled MTX in parasite cultures in parallel tetrahydrofolate much the preferred form for the plasmodial FPGS with a near-equimolar concentration of radiolabelled folinic acid activity. This is in marked contrast to the properties of human FPGS, (5-formylTHF), we found by HPLC analysis that no polyglutamy- which is able to polyglutamylate the dihydro-form of folic acid with lated versions of MTX could be detected after overnight incubation efficiencies comparable to that for the tetrahydrofolate form, and (Fig. 6a). In contrast, cultures supplemented with 3H-folinic acid can also modify the fully oxidised form to a significant, albeit lesser, generated a large amount of labelled polyglutamylated folates up degree [36,42]. to the pentaglutamate level (Fig. 6b), as seen in previous reports [8,9,38]. Indeed, the conversion of 3H-folinic acid to such forms over this period was almost complete. Moreover, when the effi- 3.6. Methotrexate as a possible substrate for PfFPGS ciency of uptake and retention of the two radiolabelled compounds was compared, MTX was found to have accumulated in the para- The above analysis showed that plasmodial FPGS is very poor sites to somewhat higher levels than folinic acid (Fig. 6c), showing at adding further glutamate residues to folic acid and thus differs that it is not the transport into the cell that limits the polyglutamy- importantly from human FPGS in this respect. This observation lation of MTX, but the lack of subsequent processing. This result led us to test whether methotrexate (MTX) could act as a sub- clearly indicated that MTX is not a substrate for the plasmodial strate for the plasmodial enzyme. MTX is a fully oxidised folic FPGS activity. acid analogue extensively used in anti-cancer antifolate therapy, To further investigate any possible interaction between MTX whose efficacy depends not only on its binding to the DHFR of and PfDHFS–FPGS, MTX was tested as a potential inhibitor in indi- rapidly dividing cancer cells, but also on its polyglutamylated vidual DHFS and FPGS assays. However, neither the DHFS nor FPGS forms binding to other enzymes in the folate pathway [45]. The activities showed any sensitivity to this drug, even at concentra- FPGS in human cells can convert MTX into such forms, which, as

Fig. 5. Substrate speciﬁcity of PfDHFS–FPGS. Glutamylation of equal concentrations of DHP, folic acid (F), DHF and THF by PfDHFS–FPGS under identical conditions: (a) 25 ␮M DHP, F, DHF or THF, 25 ␮M ATP and 18 ␮M glutamate; (b) 100 ␮M DHP, F, DHF or THF, 2 mM ATP and 200 ␮M glutamate. Glutamylation activity is represented as a percentage of that seen with DHP as substrate in each case. P. Wang et al. / Molecular & Biochemical Parasitology 172 (2010) 41–51 47

Fig. 6. MTX accumulation in, and inhibition of, P. falciparum. A predominantly ring-stage culture of FCB isolate was divided equally and each half grown for 20 h in the presence of (a) 3H-MTX or (b) 3H-folinic acid in parallel and extracts analysed by HPLC. The arrows indicate the position of MTX and the mono- and polyglutamylated forms (FPG3-5) of folinic acid. (c) Accumulation of MTX (black) in parasite cells compared with folinic acid (white). Counts in the culture supernatant (converted to nM) indicate the (equal) initial label concentrations to which each culture was exposed; ‘extract’ represents the counts extracted from the equal numbers of parasites in each pellet labelled with either MTX or folinic acid after the incubation. (d) Dose–response curves of different strains of cultured malaria parasites to increasing MTX concentration. Values are relative to growth in the absence of drug, taken as 100%. For reasons as yet unclear, extremely low levels of MTX (around 1 nM) have a stimulatory effect on growth.

tions as high as 1.0 mM (Supplementary Fig. 5a and b). In contrast, of 0.41 ␮M for the DHFS activity of the recombinant protein while growth of P. falciparum cultures was highly sensitive to MTX, with the value for the 7,8-dihydro derivative of 2 against the FPGS activ- average IC50 values of ∼40 nM for four different strains of the par- ity was 0.39 ␮M(Supplementary Fig. 6). The speciﬁcity of these asite (Fig. 6d), as seen previously with other strains [50], i.e. about two inhibitors was more rigorously studied by using Vmax values 25,000 times lower, presumably as a result of its known, very tight in secondary plots to calculate the Ki values for both compounds binding to DHFR (Ki < 1 nM), regardless of whether this enzyme car- against both activities. The data showed that reduced 1 was about ries the mutations that confer PYR resistance [51]. This indicates a 5-fold more potent in inhibiting the DHFS activity than the FPGS ␮ complete absence of either allosteric or competitive effects of MTX activity. Thus, it had a Ki value of 0.14 M for the former and one on the parasite DHFS–FPGS enzyme. We conclude that if MTX has of 0.63 ␮M for the latter. Conversely, reduced 2 was a much more more than one target in P. falciparum, as has been suggested, it is not potent inhibitor of FPGS activity (almost 20-fold), with a Ki value mediated via polyglutamylated forms of the drug, unlike in mam- of 0.091 ␮M for the FPGS activity and of 1.69 ␮M for the DHFS malian systems, and further that the DHFS and FPGS activities are activity (Fig. 8). These ﬁgures were derived under low substrate unaffected by this antifolate agent.

3.7. Inhibition of the PfDHFS–FPGS activities by two novel folate analogues

To date, the reactions catalysed by PfDHFS–FPGS have not been exploited, nor even explored, as possible targets for antimalarial inhibitors. Given that DHFS has no counterpart in mammals and that the parasite FPGS activity shows properties signiﬁcantly different from those of human FPGS, identifying effective and selective inhibitors of these activities would be an attractive goal. We therefore investigated two novel compounds that had been designed as folate analogue inhibitors by mimicking the unstable tetrahedral intermediates derived from the acyl phosphate intermediates formed during DHFS- and FPGS-catalysed ligation of glutamic acid to DHP and THF, respectively [52]. Thus, two tetrahedral phos- Fig. 7. Structures of the putative PfDHFS–FPGS inhibitors, compounds 1 and 2. These phinic acids (1 [31] and 2 [33]; Fig. 7) were evaluated as predicted are folate analogues incorporating aryl- and alkylphosphinic acid moieties [1 and 2, respectively] to simulate postulated unstable tetrahedral intermediates for the two inhibitors of PfDHFS (1) and PfFPGS (2), respectively. In exploratory activities [31,33]. They are shown here in their oxidised forms; they were reduced experiments, the 7,8-dihydro derivative of 1 exhibited an IC50 value to their 7,8-dihydropterin forms for the inhibition studies. 48 P. Wang et al. / Molecular & Biochemical Parasitology 172 (2010) 41–51

Fig. 8. Secondary plots of 1/Vmax of PfDHFS and PfFPGS activities against varying concentrations of 1 and 2 for the determination of Ki values of the inhibitors. In order to obtain Km and Vmax data (see Supplementary Fig. 7), DHP was used as the variable substrate in (a) and (c) for DHFS activity; THF in (b) and (d) for FPGS activity. ATP = 50 ␮M, glutamate = 25 ␮M. H2-1 has a Ki of 0.14 ␮M for DHFS and a Ki of 0.63 ␮M for FPGS activity. H2-2 has a Ki of 1.69 ␮M for DHFS activity and a Ki of 0.091 ␮M for FPGS activity, as indicated by arrows. concentration conditions where data fit to the Michaelis–Menten their binding sequentially to form a quaternary complex, with DHP equation was good (see above), but were little changed at high con- having the highest affinity among the three substrates. This would centrations of ATP and glutamate (data not shown), where Vmax be consistent with a low level of synthesis of DHP in the para- values were determined using the full equation of model a in sites relative to the availability of ATP and glutamate. However, Supplementary Fig. 2. we demonstrate that the plasmodial DHFS activity is an unusual To assess the mode of binding of the two drugs to PfDHFS–FPGS, enzyme type in that accurate modelling of its kinetics required the either DHP was used as the variable substrate for the DHFS activity concept of a second DHP binding site to be introduced together or THF as that for the FPGS activity with the inhibitors used in a with the assumption that the binding of this second site by DHP series of constant concentration steps in each data set to yield the inhibits the enzyme’s activity, but that such inhibition is only par- double-reciprocal plots of Supplementary Fig. 7. Despite converg- tial, even at high DHP concentrations. A similar phenomenon has ing to points in all of the plots, these lie off both the x- and y-axes, been observed for DHPS from Staphylococcus aureus with respect as both inhibitors induced changes in both the Km and the Vmax to pAB [53] and Cryptosporidium parvum inosine monophosphate values of the reaction. This indicates that the effects of the drugs dehydrogenase with respect to NAD [54]. In our experiments, sub- are not uncompetitive, but neither are they fully competitive nor strate inhibition became significant at DHP concentrations higher fully non-competitive, with both apparently exerting their effect than ∼2 ␮M and was most pronounced when the co-substrates ATP through a mixed mode of competitive action on the bifunctional and glutamate were readily available at the higher concentrations DHFS–FPGS enzyme of the parasite. expected in vivo. Thus in general, ATP concentrations are in the mM range in most cell types investigated [55], and this is also true of 4. Discussion P. falciparum blood stages, where an average level of ∼2.5 mM has been measured for parasites cultured in physiological glucose [56]. Folate biosynthesis and metabolism is a long-established clin- Similarly, studies on P. lophurae indicate that glutamate levels are ical target for antimalarial intervention. We have demonstrated normally at least 1 mM and probably higher in the cytosol [57]. previously that the malarial genome encodes a bifunctional The unusual kinetics we observe suggest that under normal in vivo DHFS–FPGS as a component of its folate pathway and verified its conditions, production of the folate moiety might be a step in the predicted roles by complementation experiments in mutants of biosynthetic pathway at which a certain degree of negative feed- yeast and E. coli defective in these activities [30,34]. However, a back is operational, depending upon the (unknown) concentrations thorough kinetic study has not yet been carried out on this proof DHP. In contrast, no evidence was seen for substrate inhibition tein, whose functions are critical to both folate synthesis and folate in the FPGS reaction with THF as substrate under any conditions, modification, from this or any other protozoan parasite. We first quite different to what is observed with human FPGS [43,44,58]. The ∼ ␮ studied the binding of the substrates of the PfDHFS activity by apparent Km values of 1 M that we determined for the parasite systematically varying the concentrations of the three compounds DHFS and FPGS activities with respect to their principal sub- involved, DHP, ATP and glutamate. Our data are consistent with strates are comparable to those seen with other homologues, e.g. P. Wang et al. / Molecular & Biochemical Parasitology 172 (2010) 41–51 49

∼0.4–6 ␮M for DHP with bacterial DHFS [22,24,37] and ∼2–5 ␮M perhaps surprising, and suggests that in this case the nature of the for THF with bacterial and mammalian FPGS [37,42,59,60]. Values heterocycle is less important for binding than is the phosphinate of Vmax were very variable, depending upon the enzyme prepara- mimic of the tetrahedral intermediate. These in vitro inhibitor stud- tion and its age, reflecting significant instability, as also reported ies on purified enzyme serve as an important proof of principle, but for a range of other DHFS/FPGS molecules, both prokaryotic and in preliminary experiments, the reduced compounds were unable eukaryotic [22,25,28,36,59]. This, together with the DHP inhibi- to significantly inhibit parasite growth in culture, probably due to tion phenomenon observed here for the PfDHFS activity made instability over long periods at 37 ◦C and/or insufficient cell perme- meaningful comparisons with values reported for other systems ability. Such analogues would therefore require further appropriate difficult. However, the highest turnover numbers we calculated structural modifications to overcome these current limitations. were 0.04 s−1 (uninhibited) and 0.01 s−1 (substrate inhibited), sig- The study using MTX as a candidate inhibitor yielded several nificantly slower than reported for human and bacterial enzymes interesting results. This major anti-cancer antifolate is reported (0.6–0.8 s−1 [25,37,42]), but similar to figures of 0.02–0.03 s−1 to inhibit multiple enzymes in mammalian cells in addition to reported for the P. falciparum HPPK–DHPS bifunctional enzyme that its traditional target of DHFR [18]. This secondary mechanism of precedes DHFS in the folate biosynthetic pathway [61,62]. Another MTX inhibition is due to its polyglutamylation by cellular FPGS enzyme in the thymidylate cycle component of the P. falciparum activity, increasing its affinity for other key molecules, including folate pathway, serine hydroxymethyltransferase (SHMT), exhibits TS and other folate-dependent enzymes [46]. Moreover, in cells faster turnover numbers (0.16–0.74 s−1 [63,64]), but are similarly that can potentially develop resistance to MTX by pumping the calculated as being, on average, about 30 times slower than their unaltered drug outside of the cell membrane, polyglutamylation human or bacterial equivalents [63]. enhances its accumulation by virtue of the increases in negative Our studies of substrate specificity also indicate that the prod- charge. Our study of the ability of plasmodial FPGS to catalyse uct of the DHFS step, DHF, despite its location at the catalytic site of MTX polyglutamylation, both in vitro and in vivo, again highlighted the DHFS activity, almost certainly must dissociate and undergo major differences between parasite and human FPGS activities. No reduction by DHFR to THF, before rebinding in this form to the polyglutamylated forms of MTX could be detected after overnight site of FPGS activity for conversion into polyglutamylated forms. cultivation with the 3H-MTX label whereas the conversion of a Whether the two active sites are discrete or overlapping in bacte- near-equimolar concentration of 3H-folinic acid, a folate with a ria with a bifunctional DHFS–FPGS has been the subject of some fully reduced (tetrahydro-) pterin ring, was almost complete under controversy [21,23,24], but our data on the parasite protein (Fig. 4) identical conditions, a phenomenon that we showed was not due are more consistent with the latter view, that DHP and THF must to poor transport and accumulation of the drug in the parasite, occupy the same or extensively overlapping catalytic sites. This in which actually exceeded that of folinic acid. In fact, we could find turn raises the question as to how the relative levels of folate pro- no evidence for any interaction between PfDHFS–FPGS and MTX, duction and subsequent polyglutamylation are controlled, if at all, nor more than a very low level of activity on folic acid, which MTX in this organism. resembles closely, especially in the fully oxidised nature of the The most obvious difference between parasite and host with pterin ring. Whether MTX inhibits any enzyme in the parasite folate regard to these aspects of folate metabolism is that, as in the pathway other than its classical target DHFR, or indeed any other case of E. coli and other bacteria, the plasmodial protein has two non-folate related activities, remains to be established, but we can functions whereas the human enzyme has only the FPGS activity confidently exclude MTX inhibition of PfDHFS–FPGS and any role and entirely lacks a DHFS, along with other enzymes of de novo of polyglutamylated forms of this drug in its powerful antimalarial folate synthesis. However, there is a significant level of similar- effect. ity between human FPGS, DHFS–FPGS from the malarial parasite Taking the above studies on phosphinic acid folate analogues and DHFS or FPGS enzymes from other organisms in their pri- and MTX together, our results emphasise that, despite the close mary protein sequences [34]. They all have an active site located similarities in their function and the likely extensive overlap in in the same area of the enzyme and the binding of ATP and glu- their catalytic site, the DHFS and FPGS activities of the parasite can tamate is conserved among all of them. Importantly though, the be differentiated from each other at the inhibitor level. Moreover, substrate specificity studies reported here demonstrate that, in the kinetic and substrate specificity experiments reveal that the addition, the parasite FPGS has properties that differ significantly parasite FPGS shows significant differences relative to its human from those reported for the orthologue of its host, particularly with orthologue. These observations, together with the absence of DHFS respect to the oxidation state of the folate to be glutamylated, activity in humans, open up the possibility of developing inhibitors further justifying the bifunctional parasite molecule as a putative against the bifunctional DHFS–FPGS protein that are highly specific novel drug target. Aryl phosphinic acid analogues of folic acid and for the parasite. several antifolates have been synthesised as potential inhibitors of DHFS [31]. With targeting of bifunctional DHFS–FPGS in mind, including that of P. falciparum, we tested for the first time such Acknowledgements inhibitors against the plasmodial target activities. Thus, using the aryl phosphinic analogue of folic acid (1), we demonstrated that We thank Abraham Louw and Esmare Human (University of Pre- this compound in its 7,8-dihydro reduced form (but not its oxi- toria) for help with preliminary experiments towards expression dised form) strongly inhibited the DHFS activity of the parasite of the pfdhfs-fpgs gene and Jeremy Derrick (University of Manch- with a Ki of 140 nM, but was about 5-fold less effective against the ester) for helpful discussions. We also thank David Bartley and John FPGS activity. We also tested an alkyl phosphinic acid analogue of Tomsho (University of Michigan) for the sample of compound 2 pteroylglutamyl-␥-glutamate (2), an inhibitor originally designed used in these experiments. We acknowledge the Wellcome Trust, to target the human FPGS activity [32,33,65] against both of the UK [grant number 073896] for support of this work. activities of PfDHFS–FPGS. Consistent with its design, this analogue, again in its reduced form, showed a marked preference for the plas- Appendix A. Supplementary data modial FPGS activity with a Ki of 91 nM, almost 20-fold more potent than that measured for the DHFS activity (1.69 ␮M). Given how poor we found DHF itself to be as a substrate for PfFPGS, the strik- Supplementary data associated with this article can be found, in ing effectiveness of the dihydro-form of 2 against this activity is the online version, at doi:10.1016/j.molbiopara.2010.03.012. 50 P. Wang et al. / Molecular & Biochemical Parasitology 172 (2010) 41–51

References [31] Yang YH, Coward JK. Synthesis of p-aminophenyl aryl h-phosphinic acids and esters via cross-coupling reactions: elaboration to phosphinic acid pseudopep- [1] Breman JG. The ears of the hippopotamus: manifestations, determinants, and tide analogues of pteroyl glutamic acid and related antifolates. J Org Chem estimates of the malaria burden. Am J Trop Med Hyg 2001;64:1–11. 2007;72:5748–58. [2] Suh KN, Kain KC, Keystone JS. Malaria. Can Med Ass J 2004;170:1693–702. [32] Bartley DM, Coward JK. A stereoselective synthesis of phosphinic acid phos- [3] Müller IB, Hyde JE, Wrenger C. Vitamin B metabolism in Plasmodium falciparum phapeptides corresponding to glutamyl-gamma-glutamate and incorporation as a source of drug targets. Trends Parasitol 2010;26:35–43. into potent inhibitors of folylpoly-gamma-glutamyl synthetase. J Org Chem [4] McGuire JJ, Coward JK. Pteroylpolyglutamates: biosynthesis, degradation, and 2005;70:6757–74. function. In: Blakley RL, Benkovic SJ, editors. Folates and pterins: Chemistry [33] McGuire JJ, Bartley DM, Tomsho JW, Haile WH, Coward JK. Inhibition of human and biochemistry of folates. New York: Wiley; 1984. p. 135–90. folylpolyglutamate synthetase by diastereomeric phosphinic acid mimics of [5] Shane B. Folylpolyglutamate synthesis and role in the regulation of one-carbon the tetrahedral intermediate. Arch Biochem Biophys 2009;488:140–5. metabolism. Vitam Horm Adv Res Appl 1989;45:263–335. [34] Lee CS, Salcedo E, Wang Q, Wang P, Sims PFG, Hyde JE. Characterization of three [6] Osborne CB, Lowe KE, Shane B. Regulation of folate and one-carbon metabolism genes encoding enzymes of the folate biosynthetic pathway in Plasmodium in mammalian cells. 1. Folate metabolism in Chinese hamster ovary cells falciparum. Parasitology 2001;122:1–13. expressing Escherichia coli or human folylpoly-gamma-glutamate synthetase [35] Studier FW. Protein production by auto-induction in high-density shaking cul- activity. J Biol Chem 1993;268:21657–64. tures. Protein Expr Purif 2005;41:207–34. [7] Toy J, Bognar AL. Mutagenesis of the Lactobacillus casei folylpolyglutamate [36] McGuire JJ, Hsieh P, Coward JK, Bertino JR. Enzymatic synthesis of synthetase gene at essential residues resembling an ATP binding-site. Arch folylpolyglutamates—characterization of the reaction and its products. J Biol Biochem Biophys 1994;314:344–50. Chem 1980;255:5776–88. [8] Krungkrai J, Webster HK, Yuthavong Y. De novo and salvage biosynthe- [37] Bognar AL, Shane B. Bacterial folylpoly(gamma-glutamate) synthase– sis of pteroylpentaglutamates in the human malaria parasite, Plasmodium dihydrofolate synthase. Methods Enzymol 1986;122:349–59. falciparum. Mol Biochem Parasitol 1989;32:25–37. [38] Wang P, Nirmalan N, Wang Q, Sims PFG, Hyde JE. Genetic and metabolic analysis [9] Wang P, Wang Q, Aspinall TV, Sims PFG, Hyde JE. Transfection studies to explore of folate salvage in the human malaria parasite Plasmodium falciparum. Mol essential folate metabolism and antifolate drug synergy in the human malaria Biochem Parasitol 2004;135:77–87. parasite Plasmodium falciparum. Mol Microbiol 2004;51:1425–38. [39] Pastore EJ, Friedkin M, Jardetzky O. Structure of dihydrofolic acid prepared by [10] McBurney MW, Whitmore GF. Isolation and biochemical characterization of dithionite reduction of folic acid. J Am Chem Soc 1963;85:3058–9. folate deficient mutants of Chinese hamster cells. Cell 1974;2:173–82. [40] Hillcoat BL, Blakley RL. Reduction of folate by borohydride and by dithionite. [11] Taylor RT, Hanna ML. Folate-dependent enzymes in cultured Chinese hamster Biochem Biophys Res Commun 1964;15:303–7. cells—folylpolyglutamate synthetase and its absence in mutants auxotrophic [41] Fang LJ, King J. Purification and characterization of folylpolyglutamate for glycine + adenosine + thymidine. Arch Biochem Biophys 1977;181:331–44. synthetase (FPGS) from cultured cells of Datura innoxia. J Plant Physiol [12] Cherest H, Thomas D, Surdin-Kerjan Y. Polyglutamation of folate coenzymes is 1997;151:528–34. necessary for methionine biosynthesis and maintenance of intact mitochon- [42] Atkinson I, Garrow T, Brenner A, Shane B. Human cytosolic folylpoly-gamma- drial genome in Saccharomyces cerevisiae. J Biol Chem 2000;275:14056–63. glutamate synthase. Methods Enzymol 1997;281:134–40. [13] Schirch V, Strong WB. Interaction of folylpolyglutamates with enzymes in one- [43] Moran RG, Baldwin SW, Taylor EC, Shih C. The 6S-diastereomer and carbon metabolism. Arch Biochem Biophys 1989;269:371–80. 6R-diastereomer of 5,10-dideaza-5,6,7,8-tetrahydrofolate are equiac- [14] Besson V, Rebeille F, Neuburger M, Douce R, Cossins EA. Effects of tive inhibitors of denovo purine synthesis. J Biol Chem 1989;264: tetrahydrofolate polyglutamates on the kinetic parameters of serine hydrox- 21047–51. ymethyltransferase and glycine decarboxylase from pea leaf mitochondria. [44] Tomsho JW, McGuire JJ, Coward JK. Synthesis of (6R)- and (6S)-5,10- Biochem J 1993;292(Pt 2):425–30. dideazatetrahydrofolate oligo-gamma-glutamates: kinetics of multiple glu- [15] Krumdieck CL, Eto I, Baggott JE. Regulatory role of oxidized and reduced tamate ligations catalyzed by folylpoly-gamma-glutamate synthetase. Org pteroylpolyglutamates. Ann NY Acad Sci 1992;669:44–58. Biomol Chem 2005;3:3388–98. [16] Sybesma W, Van Den Born E, Starrenburg M, et al. Controlled modulation of [45] Kim JS, Lowe KE, Shane B. Regulation of folate and one-carbon metabolism folate polyglutamyl tail length by metabolic engineering of Lactococcus lactis. in mammalian cells. 4. Role of folylpoly-gamma-glutamate synthetase Appl Environ Microbiol 2003;69:7101–7. in methotrexate metabolism and cytotoxicity. J Biol Chem 1993;268: [17] Galivan J, Inglese J, McGuire JJ, Nimec Z, Coward JK. Gamma- 21680–5. fluoromethotrexate—synthesis and biological activity of a potent inhibitor [46] Synold TW, Willits EM, Barredo JC. Role of folylpolygutamate synthetase (FPGS) of dihydrofolate reductase with greatly diminished ability to form in antifolate chemotherapy; a biochemical and clinical update. Leuk Lymphoma poly-gamma-glutamates. Proc Natl Acad Sci USA 1985;82:2598–602. 1996;21:9–15. [18] Barnes MJ, Estlin EJ, Taylor GA, et al. Impact of polyglutamation on sensitivity [47] Allegra CJ, Chabner BA, Drake JC, Lutz R, Rodbard D, Jolivet J. Enhanced inhi- to raltitrexed and methotrexate in relation to drug-induced inhibition of de bition of thymidylate synthase by methotrexate polyglutamates. J Biol Chem novo thymidylate and purine biosynthesis in CCRF-CEM cell lines. Clin Cancer 1985;260:9720–6. Res 1999;5:2548–58. [48] Chabner BA, Allegra CJ, Curt GA, et al. Polyglutamation of methotrexate. Is [19] Shane B. Pteroylpoly(gamma-glutamate) synthesis by Corynebacterium methotrexate a prodrug? J Clin Invest 1985;76:907–12. species: in vivo synthesis of folates. J Biol Chem 1980;255:5649–54. [49] Nduati E, Diriye A, Ommeh S, et al. Effect of folate derivatives on the [20] Fussenegger M, Meyer TF. Cloning and characterization of the Neisseria gonor- activity of antifolate drugs used against malaria and cancer. Parasitol Res rhoeae MS11 folC gene. Mol Gen Genetics 1996;250:277–85. 2008;102:1227–34. [21] Young PG, Smith CA, Metcalf P, Baker EN. Structures of Mycobacterium tuber- [50] Fidock DA, Wellems TE. Transformation with human dihydrofolate reductase culosis folylpolyglutamate synthase complexed with ADP and AMPPCP. Acta renders malaria parasites insensitive to WR99210 but does not affect the intrin- Crystallogr D Biol Crystallogr 2008;D64:745–53. sic activity of proguanil. Proc Natl Acad Sci USA 1997;94:10931–6. [22] Bognar AL, Osborne C, Shane B, Singer SC, Ferone R. Folylpoly-gamma- [51] Chen GX, Mueller C, Wendlinger M, Zolg JW. Kinetic and molecular prop- glutamate synthetase-dihydrofolate synthetase—cloning and high expression erties of the dihydrofolate reductase from pyrimethamine-sensitive and of the Escherichia coli folC gene and purification and properties of the gene pyrimethamine-resistant clones of the human malaria parasite Plasmodium product. J Biol Chem 1985;260:5625–30. falciparum. Mol Pharmacol 1987;31:430–7. [23] Mathieu M, Debousker G, Vincent S, Viviani F, Bamas-Jacques N, Mikol V. [52] Banerjee RV, Shane B, McGuire JJ, Coward JK. Dihydrofolate synthetase and Escherichia coli FolC structure reveals an unexpected dihydrofolate binding folylpolyglutamate synthetase—direct evidence for intervention of acyl phos- site providing an attractive target for anti-microbial therapy. J Biol Chem phate intermediates. Biochemistry 1988;27:9062–70. 2005;280:18916–22. [53] Hampele IC, Darcy A, Dale GE, et al. Structure and function of the dihy- [24] Sheng Y, Khanam N, Tsaksis Y, Shi XM, Lu QS, Bognar AL. Mutagenesis of dropteroate synthase from Staphylococcus aureus. J Mol Biol 1997;268: folylpolyglutamate synthetase indicates that dihydropteroate and tetrahydro- 21–30. folate bind to the same site. Biochemistry 2008;47:2388–96. [54] Umejiego NN, Li C, Riera T, Hedstrom L, Striepen B. Cryptosporidium parvum IMP [25] Bognar AL, Shane B. Purification and properties of Lactobacillus casei folylpoly- dehydrogenase—identification of functional, structural, and dynamic proper- gamma-glutamate synthetase. J Biol Chem 1983;258:2574–81. ties that can be exploited for drug design. J Biol Chem 2004;279:40320–7. [26] Cichowicz DJ, Shane B. Mammalian folylpoly-gamma-glutamate synthetase. 2. [55] Gribble FM, Loussouarn G, Tucker SJ, Zhao C, Nichols CG, Ashcroft FM. A novel Substrate specificity and kinetic properties. Biochemistry 1987;26:513–21. method for measurement of submembrane ATP concentration. J Biol Chem [27] Sun X, Bognar AL, Baker EN, Smith CA. Structural homologies with ATP- and 2000;275:30046–9. folate-binding enzymes in the crystal structure of folylpolyglutamate syn- [56] Saliba KJ, Kirk K. pH regulation in the intracellular malaria parasite, Plas- thetase. Proc Natl Acad Sci USA 1998;95:6647–52. modium falciparum—H+ extrusion via a V-type H+-ATPase. J Biol Chem [28] McDonald D, Atkinson IJ, Cossins EA, Shane B. Isolation of dihydrofolate 1999;274:33213–9. and folylpolyglutamate synthetase activities from Neurospora. Phytochemistry [57] Sherman IW, Mudd JB. Malaria infection (Plasmodium lophurae): changes in free 1995;38:327–33. amino acids. Science 1966;154:287–9. [29] Cossins EA, Chen LF. Folates and one-carbon metabolism in plants and fungi. [58] Clarke L, Waxman DJ. Human liver folylpolyglutamate synthetase— Phytochemistry 1997;45:437–52. biochemical characterization and interactions with folates and folate [30] Salcedo E, Cortese JF, Plowe CV, Sims PFG, Hyde JE. A bifunctional dihydrofolate antagonists. Arch Biochem Biophys 1987;256:585–96. synthetase–folylpolyglutamate synthetase in Plasmodium falciparum identified [59] Shane B. Pteroylpoly(gamma-glutamate) synthesis by Corynebacterium by functional complementation in yeast and bacteria. Mol Biochem Parasitol species—purification and properties of folylpoly(gamma-glutamate) syn- 2001;112:241–54. thetase. J Biol Chem 1980;255:5655–62. P. Wang et al. / Molecular & Biochemical Parasitology 172 (2010) 41–51 51

[60] Chen L, Qi H, Korenberg J, Garrow TA, Choi YJ, Shane B. Puriﬁcation and [63] Pang CKT, Hunter JH, Gujjar R, et al. Catalytic and ligand-binding characteris- properties of human cytosolic folylpoly-gamma-glutamate synthetase and tics of Plasmodium falciparum serine hydroxymethyltransferase. Mol Biochem organization, localization, and differential splicing of its gene. J Biol Chem Parasitol 2009;168:74–83. 1996;271:13077–87. [64] Maenpuen S, Sopitthummakhun K, Yuthavong Y, Chaiyen P, Leart- [61] Triglia T, Menting JGT, Wilson C, Cowman AF. Mutations in dihydropteroate sakulpanich U. Characterization of Plasmodium falciparum serine synthase are responsible for sulfone and sulfonamide resistance in Plasmodium hydroxymethyltransferase—a potential antimalarial target. Mol Biochem falciparum. Proc Natl Acad Sci USA 1997;94:13944–9. Parasitol 2009;168:63–73. [62] Kasekarn W, Sirawaraporn R, Chahomchuen T, Cowman AF, Sirawaraporn [65] Valiaeva N, Bartley D, Konno T, Coward JK. Phosphinic acid pseudopeptides W. Molecular characterization of bifunctional hydroxymethyldihydropterin analogous to glutamyl-gamma-glutamate: synthesis and coupling to pteroyl pyrophosphokinase–dihydropteroate synthase from Plasmodium falciparum. azides leads to potent inhibitors of folylpoly-gamma-glutamate synthetase. J Mol Biochem Parasitol 2004;137:43–53. Org Chem 2001;66:5146–54.