Transfection studies to explore essential folate metabolism and antifolate drug synergy in the human malaria parasite Plasmodium falciparum

Authors

  • Ping Wang,

    1. Department of Biomolecular Sciences, University of Manchester Institute of Science and Technology (UMIST), PO Box 88, Manchester M60 1QD, UK.
    Search for more papers by this author
  • Qi Wang,

    1. Department of Biomolecular Sciences, University of Manchester Institute of Science and Technology (UMIST), PO Box 88, Manchester M60 1QD, UK.
    Search for more papers by this author
  • Tanya V. Aspinall,

    1. Department of Biomolecular Sciences, University of Manchester Institute of Science and Technology (UMIST), PO Box 88, Manchester M60 1QD, UK.
    Search for more papers by this author
  • Paul F. G. Sims,

    1. Department of Biomolecular Sciences, University of Manchester Institute of Science and Technology (UMIST), PO Box 88, Manchester M60 1QD, UK.
    Search for more papers by this author
  • John E. Hyde

    Corresponding author
    1. Department of Biomolecular Sciences, University of Manchester Institute of Science and Technology (UMIST), PO Box 88, Manchester M60 1QD, UK.
      E-mail john.hyde@umist.ac.uk; Tel. (+44) 161 200 4185; Fax (+44) 161 236 0409.
    Search for more papers by this author

E-mail john.hyde@umist.ac.uk; Tel. (+44) 161 200 4185; Fax (+44) 161 236 0409.

Summary

Folate metabolism in Plasmodium falciparum is the target of important antimalarial agents. The biosynthetic pathway converts GTP to polyglutamated derivatives of tetrahydrofolate (THF), essential cofactors for DNA synthesis. Tetrahydrofolate can also be acquired by salvage mechanisms. Using a transfection system adapted to studying this pathway, we investigated modulation of dihydropteroate synthase (DHPS) activity on parasite phenotypes. Dihydropteroate synthase incorporates p-aminobenzoate (pABA) into dihydropteroate, the precursor of dihydrofolate. We were unable to obtain viable parasites where the dhps gene had been truncated. However, parasites where the protein was full-length but mutated at two key residues and having < 10% of normal activity were viable in folate-supplemented medium. Metabolic labelling showed that these parasites could still convert pABA to polyglutamated folates, albeit at a very low level, but they could not survive on pABA supplementation alone. This degree of disablement in DHPS also abolished the synergy of the antifolate combination pyrimethamine/sulfadoxine. These data indicate that DHPS activity above a low but critical level is essential regardless of the availability of salvageable folate and formally prove the role of this enzyme in antifolate drug synergy and folate biosynthesis in vivo. However, we found no evidence of a significant role for DHPS in folate salvage. Moreover, when biosynthesis was compromised by the absence of a fully functional DHPS, the parasite was able to compensate by increasing flux through the salvage pathway.

Introduction

Chemotherapeutic agents targeting the folate metabolic pathway have long been of major clinical importance in combating bacterial and protozoan pathogens, including the lethal species of the malaria parasite of humans, Plasmodium falciparum. Drugs such as sulfadoxine (SDX) and pyrimethamine (PYR) are widely used against strains of chloroquine-resistant P. falciparum and target two enzymes in the pathway, dihydropteroate synthase (DHPS) and dihydrofolate reductase (DHFR) respectively (Sibley et al., 2001). Inhibition of folate metabolism disrupts the supply of tetrahydrofolate (THF) cofactors essential for one-carbon transfer reactions, including the formation of dTMP needed for DNA synthesis. As well as the (bifunctional) genes encoding the DHPS and DHFR activities (pppk-dhps; Brooks et al., 1994; Triglia and Cowman, 1994; and dhfr-ts; Bzik et al., 1987), genes encoding other enzymes in the folate pathway have recently been characterized (Alfadhli and Rathod, 2000; Lee et al., 2001; Salcedo et al., 2001), which may present potential new targets for antimalarial agents in the face of increasing levels of resistance to current antifolate drugs.

Biosynthesis of the folate moiety utilizes as precursors GTP, p-aminobenzoic acid (pABA) and glutamate, and is mediated by five enzyme activities, whereas interconversion of folate among the various forms required for 1-carbon transfer reactions is principally carried out by a further four activities, one of which, FPGS, converts reduced folates to polyglutamated forms. This is critical for their cellular retention as well as enhancing their affinity for other folate-dependent enzymes (Krumdieck et al., 1992). However, unlike many microorganisms, P. falciparum can salvage folate as well as synthesize it de novo (Krungkrai et al., 1989) and a key question, with important implications for antifolate therapy, concerns the relative importance of the biosynthetic pathway compared to folate salvage from the human host in vivo. Inhibition studies of the DHPS activity with high levels of SDX indicate that, at least in certain strains grown in vitro, blockage of folate biosynthesis can apparently be bypassed via the salvage route, suggesting that biosynthesis may be dispensable (Wang et al., 1997b). In the field, however, a strong correlation has been found between SDX-resistant forms of the parasite carrying mutations in the dhps gene and the usage of Fansidar, the clinical formulation of SDX and PYR (e.g. Wang et al., 1997a; Plowe et al., 1998; Nzila et al., 2000; Sibley et al., 2001), suggesting that in normal infections of the human host, folate salvage cannot completely satisfy the parasite's requirements. We have proposed that the observed synergy between SDX and PYR may result from the ability of PYR to interfere with the efficient utilization of such salvaged folate (Sims et al., 1999; Wang et al., 1999).

To investigate the relative importance of salvage and biosynthesis further, we have exploited here a gene disruption protocol that we recently developed for studying folate metabolism in this organism (Wang et al., 2002). In this, the sequence used for modification is positively selected with antibiotic markers that, unlike those normally used in parasite transfection (e.g. Crabb et al., 1997; Fidock et al., 1998) are not active against the enzymes of the folate pathway. We have initially targeted the dhps domain of pppk-dhps, the bifunctional gene whose products hydroxymethylpterin pyrophosphokinase (PPPK) and DHPS catalyse successive steps in the formation of dihydropteroate, the immediate precursor of dihydrofolate. This gene thus constitutes a critical component of the biosynthetic pathway. As the parasite is haploid throughout most of its life cycle, including the asexual blood stages in the human host, knockout of an essential gene would be lethal, and no regulatable transfection system for P. falciparum is yet available that would allow the gene to be selectively repressed or switched off only after successful transfection. However, we reasoned that, given the ability of the parasite to salvage intact folate, it might be possible to produce viable transfectants with a disrupted gene in the biosynthetic pathway, if folate provided in the culture medium were able to satisfy all of the parasite's needs for the various folate cofactors. Such transfectants could then be used to further explore the respective roles of the biosynthetic and salvage pathways. We describe here experiments to produce pppk-dhps transfectants where the dhps domain has been disabled to varying degrees of severity, and provide evidence for the essential nature of this gene. Where stable transfection was achieved, we monitored growth characteristics of the resulting parasites, their ability to process exogenous radiolabelled precursors to polyglutamated folate end-products and their responses to antifolate drug challenge.

Results

Constructs for knockout or modification of the pppk-dhps gene

The gene knockout system we have developed uses a bsd marker to select for parasites that have taken up the transfection plasmid and then depends on integration of this plasmid at the target gene locus to enable the selectable marker, neo, to be transcribed from the endogenous promoter as a fusion product with the gene of interest. This allows positive selection for the desired integrant with G418 (Wang et al., 2002). For the pppk-dhps gene, five constructs were devised (Fig. 1 and Table 1), two of which were controls. The positive control (pPKDSneoII) yields a full-length, active version of the pppk-dhps gene product after integration, but with two amino acid changes from wild-type DHPS (as carried by the transfection host FCB) and the C-terminal fusion to neomycin phosphotransferase. This form of DHPS (A437G, A581G; K1 type) is seen in many sulfonamide-resistant parasite samples (Wang et al., 1997a) and was first characterized in the K1 strain (Brooks et al., 1994; Triglia and Cowman, 1994). The codon changes assist in confirming that the modified gene has integrated at the homologous target site. The negative control (pPKDS) lacked the neo gene in the plasmid and therefore gives blasticidin (BS)-resistant transients, but does not survive subsequent G418 selection (Wang et al., 2002).

Figure 1.

Schematic representation of an integration event.
A. The plasmid construct containing the specific targeting sequence, pppk-dhps, truncated at the 5′ end (PKDS), a marker gene (bsd) to select the parasite population harbouring the plasmid, and a marker gene (neo) to select for integrants.
B. Genomic copy of pppk-dhps before transfection.
C. The modified genome with the integrated plasmid after single crossover recombination. Primers C547 and C548 monitor the presence of plasmid sequence in any context, C575 and C511 monitor the integration event.
D. Versions of the PPPK-DHPS-NEO cassette used for transfection (see Table 1 for additional details). Asterisks indicate the positions of mutations relative to the wild-type sequence of the FCB host and arrows the deletions in the DHPS domain.

Table 1. . Plasmid constructs used for transfection of FCB and characteristics of integrants.
Plasmid constructapppk-dhps after integrationIntegrant namebPurposeReference
  • a

    . (Letter) refers to constructs illustrated in Fig. 1D.

  • b

    . Names are given in brackets where integrants did not emerge from the selection process.

(a) pPKDSlacks neo sequence; integration not selectableN/Apositive control for transfection
negative control for integration
Wang et al. (2002)
(b) pPKDSneoIIfull-length, K1-type sequence fused to neoPKDSIIpositive control for integrationWang et al. (2002)
(c) pPKDSneoT1K1-type sequence truncated by 531 bp at 3′ end, fused to neo(PKDST1)test of dhps functionthis work
(d) pPKDSneoT2K1-type sequence truncated by 65 bp at 3′ end, fused to neo(PKDST2)test of dhps functionthis work
(e) pPKDSneoM2full-length K1-type sequence with
R686Q and H688Q codon changes,
fused to neo
PKDSM2test of dhps functionthis work

The three test versions of pppk-dhps were compromised in different ways. Construct pPKDSneoM2 was also designed to yield a fusion product carrying full-length PPPK-DHPS upon integration and expression, but with the critical residues Arg-686 and His-688 in the highly conserved motif RVHDV both mutated to Gln. From model building studies of the Plasmodium molecule, based on the crystal structures of DHPS from several bacterial sources (Achari et al., 1997; Baca et al., 2000), both of these completely conserved residues are predicted to contact the pterin pyrophosphate substrate and participate in the catalytic step. As an experimental test of their importance, we introduced the R to Q and H to Q mutations into the closely related PPPK-DHPS molecule from Toxoplasma gondii (Pashley et al., 1997), where we have an E. coli-based expression system producing high levels of active enzyme (Aspinall et al., 2002). The specific activity of the mutant enzyme was severely reduced by these changes, but not eliminated entirely (mean of 7.7% relative to wild-type from three determinations; data not shown). Construct pPKDSneoT2 had a deletion at the 3′ end of 65 bp (including the stop codon) to remove the final 21 amino acid residues of the molecule including the above-mentioned RVHDV motif, together with the C-terminal stretch of alpha-helix (alpha-H; Achari et al., 1997, or alpha-8; Baca et al., 2000) that forms part of the dimer interface in bacterial systems. Construct pPKDSneoT1 was more extensively truncated from the 3′ end, with 531 bp of coding sequence removed (including the stop codon). This is equivalent to 176 amino acids or about 54% of the entire dhps domain, and thus would be expected to encode a product completely devoid of DHPS activity.

Disruption of the native pppk-dhps gene

The five constructs described were electroporated into host cells of the parasite line FCB in parallel. Transient transfectants were selected on BS and integrants on G418 as described (Wang et al., 2002). After the switch to G418 selection, drug-resistant parasites emerged after ∼10–12 days for both pPKDSneoII and pPKDSneoM2, although the latter could only be obtained when cultures were supplemented with folinic acid, a fully reduced source of exogenous folate that is more readily utilized by the parasite than folic acid (see below). However, although plasmids were successfully introduced and propagated in the parasite during BS selection, no integrants were obtained either for the pPKDS construct lacking the neo fusion (the negative control), nor, despite repeated attempts, for the two 3′-truncated constructs pPKDSneoT1 and pPKDSneoT2, after G418 selection periods of> 40 days. These were unsuccessful even in the constant presence of folinic acid.

Polymerase chain reaction was initially used to monitor the transfection and integration events as described in Figs 1 and 2. To analyse successful integration in more detail, Southern analysis of the transfectants was performed using a probe derived from the K1 pppk-dhps gene. This showed that the 3.7 kb ClaI-PstI fragment (labelled as G) from the host FCB, which carries the entire pppk-dhps sequence, disappeared in the mutants PKDSII and PKDSM2 after integration (see Table 1 for nomenclature of transfected lines) and was replaced by two new fragments of 3.0 and 2.8 kb (Fig. 2B), as predicted from the restriction sites in the transfection plasmids. This confirmed that the original pppk-dhps gene had been disrupted in the desired manner and thus any DHPS activity should be encoded from the newly integrated mutant pppk-dhps gene. To demonstrate that this was the case, RT-PCR analysis of the transfectants was carried out (Fig. 2C). The fusion product of pppk-dhps-neo could be identified from the transfected cultures using primers C073 and C575 (Wang et al., 2002), which gave no PCR signal when used with RNA prepared from the host FCB. The single band detected from PKDSII and PKDSM2 was a fragment of 1009 bp in length as predicted from the fused cDNA sequences of pppk-dhps and neo. Direct sequencing of the RT-PCR products confirmed (i) that in PKDSII the wild-type FCB dhps sequence had been replaced as the active gene with that from K1 present in the transfection plasmid (two mutant positions in codons 437 and 581), (ii) that in the PKDSM2 transcript, the additional alterations (encoding R686Q and H688Q) were present, and (iii) that in both cases correct excision of the intron from the dhps domain had occurred.

Figure 2.

A. Analysis of genomic DNA from the knockout mutants of PPPK-DHPS by PCR. Panel (a), signals generated by primers C575 and C511, diagnostic for integration; panel (b), controls for the quality of the genomic DNA preparation (primers C498 and C511); panel (c), signals from the bsd gene to monitor the presence of the plasmid in the transformed culture (primers C547 and C548). Templates are: 1, genomic DNA from transfectant PKDSII; 2, genomic DNA from transfectant PKDSM2; 3, genomic DNA from host FCB; 4, pPKDSneoII plasmid DNA; 5, pPKDSneoM2 plasmid DNA.
B. Southern blot analysis of the transfectants for integration of pPKDSneoII and pPKDSneoM2 into the host FCB genome. Total DNA extracted from parasite cultures was digested with ClaI and PstI and blots probed with a pppk-dhps fragment labelled with DIG. Lane 1, cut plasmid pPKDSneoII control; lane 2, genomic DNA from PKDSII; lane 3, genomic DNA from host FCB; lane 4, genomic DNA from PKDSM2; Lane 5, cut plasmid pPKDSneoM2 control. G indicates the genomic signal from the host FCB; P indicates the signal from the plasmid used in the transformation; I indicates the new signals generated after the integration event. Plasmid signals in lanes 2 and 4 persisted after the removal of G418 selection and are presumed to arise from the formation of stable concatamers of the plasmid as described in O’Donnell et al., 2001.
C. Transcription signals from the transfectants. M, DNA ladder, 100 bp; 1, control PCR using RNA as template from host FCB, without reverse transcription; 2, RT-PCR from FCB RNA, using oligo-dT to prime the RT step, and primers C073 and C575 for the PCR (Wang et al., 2002); 3, control PCR using RNA from PKDSII, without reverse transcription; 4, RT-PCR from PKDSII RNA; 5, PCR using RNA from PKDSM2, without reverse transcription; 6, RT-PCR from PKDSM2 RNA. All patterns were obtained using nucleic acid samples extracted  1 week after the first appearance of parasites following G418 application and remained stable in all subsequent preparations.

Growth characteristics of transfectants in varying conditions of folate supplementation

To assess how the mutant parasites responded to exogenous folate or folate precursor, the nutrient requirements of the transfectants PKDSII and PKDSM2 as well as the host FCB were studied by following the cumulative growth of these lines over a period of 15 days. This was done in custom RPMI 1640 medium devoid of pABA and folate to which either folic acid, folinic acid or pABA was individually added back as a supplement. All three parasite lines grew at a comparable rate when either folate or folinic acid was provided. However, in contrast to the host FCB and PKDSII lines, transfectant PKDSM2 with the RVH to QVQ mutations in the DHPS domain did not survive in medium where pABA was the only supplement (Fig. 3). Interestingly, cultures of this line replicated apparently normally until about day 7 after switching to the pABA-only medium, at which point they rapidly died. This indicated that the reserve of folate available to the parasite under these conditions is significant and appears to be carried over by the parasite into successive reinvasions. We note also that, although supplementation with folinic acid was necessary to obtain the PKDSM2 line following initial transfection, once established, the parasite population grew normally on folic acid. This may reflect the more demanding nutritional requirements of one or a few transfected cells emerging from the selection procedure compared to an amplified population.

Figure 3.

Growth characteristics of the host (A) and transfected lines (B and C) in different media. FL, F and P indicate that the parasites have been cultured in folate/pABA-free RPMI 1640 supplemented either with folinic acid, folic acid or pABA at 1 µg ml−1 respectively. The parasite number per microlitre is calculated by multiplying the observed number on a given day by the dilution factor(s) used during the culture process up to that point.

Uptake and processing of pABA by the transfectants

To assess the impact on folate biosynthesis of compromising the dhps gene, cultures of FCB, PKDSII and PKDSM2 were radiolabelled with 14C-pABA to monitor its conversion to folate end products, which were analysed by quantitative HPLC. Synchronized cultures of each parasite were subdivided and labelled in parallel so that the batch of blood cells used and other experimental conditions were identical. In addition, subaliquots of each culture were labelled simultaneously with 3H-hypoxanthine to measure the growth rates of each parasite line over the test period and ensure that they were closely similar. Comparisons among the parasite lines were made by calculating ratios of labelled folate levels to those of hypoxanthine uptake and normalizing to the values obtained for the FCB host. The HPLC column was calibrated using unlabelled folinic acid and the pteroylmono- to pentaglutamates (PteGlun) using a gradient allowing good separation of these species (Fig. 4A). However, products of folate metabolism extracted from parasites will not coincide exactly with these standards as the latter (with the exception of folinic acid) are all fully oxidised pteroates, whereas the former will be reduced pteroates carrying various modifications at the 5/10 positions. Thus folinic acid has a 5-formyl group, and processed intermediates and final labelled products will potentially include 5,10-methenylTHF, 5,10-methyleneTHF (the substrate of serine hydroxymethyltransferase) and 5-methylTHF (if the reported methyleneTHF reductase; Asawamahasakda and Yuthavong, 1993; is active under our culture conditions). In a study where> 40 folate derivatives of this type were synthesized (and whose HPLC gradient conditions we adopted), these modifications were in general found to reduce the retention time to a small degree (1–2 min) relative to their PteGlun counterparts, the differences diminishing as the number of glutamates increased. However, all derivatives of a particular pterin varying in oxidation level and/or modification at the 5/10 positions but carrying a given number of glutamate residues elute in a separate cluster (Selhub, 1989). Thus although it was not possible to determine the precise modifications on the pteridine ring of the products we saw, the processing of the labelled precursors to the polyglutamated forms of reduced folate was a sufficient indicator of metabolic activity for several important comparisons to be made. Typical profiles are shown in Fig. 4B and C, and quantitative analyses summarized in Table 2. In general, our system resolved at least four clusters of chromatographic products, which from their positions relative to the PteGlun standards, we could ascribe, respectively, to monoglutamates, diglutamates, triglutamates and tetra/pentaglutamates, produced in varying ratios.

Figure 4.

A. Calibration of the HPLC system for separation of folate products. Unlabelled standards were added to parasite extracts before chromatography and monitored at 280 nm; a is folinic acid and b-f are the pteroyl glutamates PteGlu1–5, respectively (n = 1 for folic acid).
B and C. Metabolic labelling of large-scale parasite cultures FCB (upper panels), PKDSII (middle panels) and PKDSM2 (lower panels). Parasites were incubated with the labels shown and the labelled products extracted, purified on a folate binding protein column and then separated on a C18 reverse-phase column. For greater sensitivity, all of the fractions from the HPLC were collected and counted in a scintillation counter. From calibrations as in (A), the peaks were assigned to clusters of monoglutamated, diglutamated, triglutamated and tetra/pentaglutamated products, as shown in the upper right panel, after taking into account the 1 min delay between the UV detector and the arrival of sample at the fraction collector. The y-axes represent cpm values adjusted for small differences in growth rates among the different parasite lines over the labelling period. The x-axes represent the cumulative volume (ml) of the fractions eluting from the HPLC column at a flow rate of 1 ml min−1.

Table 2. . Metabolic labelling of host and transfectant parasite cultures.a
Parasite linefolinic acid:folic acidbtetra/pentaglutamated:
monoglutamated productsc
pABA incorporationd
large-scalesmall-scaleefolinic acidfolic acidsmall-scale
  • a

    . Figures shown are means of values determined in triplicate and corrected for small variations in growth over the labelling period as measured by 3H-hypoxanthine incorporation (see text). Standard deviations were ≤6.4% of the mean for all values.

  • b. 

    b. Ratio of radiolabel incorporated into parallel cultures.

  • c. 

    c. Ratios calculated from the area under the relevant peaks after HPLC separation of folate products extracted from large-scale cultures using the labels shown.

  • d. 

    d. Relative values normalized to FCB.

  • e. 

    e. For definitions of scale, see text.

FCB4.213.605.100.201.00
PKDSII2.452.686.811.670.23
PKDSM23.733.759.601.740.06

It was found that all three parasite lines were able to convert pABA into folate derivatives with the highest level of polyglutamation, but to greatly differing degrees. Strikingly, it was observed that the mutated DHPS in PKDSM2 had lost virtually all of its ability to perform this conversion (Fig. 4B and Table 2), with only a very low level of activity above background still detectable. Thus, total folate product derived from labelled pABA was measured at 6% of that seen for FCB. We observed that the PKDSII line was also less efficient than the untransformed host at utilizing pABA, although the reduction here (to 23% of the FCB value) was less extreme than for PKDSM2, consistent with the ability of PKDSII to survive in medium supplemented only with pABA (Fig. 3). This difference between PKDSII and FCB is probably due to a combination of two factors; (i) although the DHPS domain in PKDSII is full-length, it is fused to neomycin phosphotransferase at its C-terminus, which could adversely affect its efficiency, and (ii), the Km for pABA for the wild-type DHPS sequence in FCB is 16 nM, whereas the PKDSII line carries the K1-type drug-resistance mutations (A437G, A581G), which increase the Km nearly fivefold to 77 nM (Triglia et al., 1997). Taken together, the data for PKDSM2 and PKDSII indicate that while DHPS need not be fully active for survival, it must exceed a certain threshold level, which appears from these experiments to be quite low.

Uptake and processing of folic acid and folinic acid by the transfectants

Whereas DHPS has a well defined role in folate biosynthesis that is clearly demonstrated by the labelling experiments with pABA above, it has been proposed that it might also participate in folate salvage. Thus, folate acquired in this way could be broken down into its pterin aldehyde and p-aminobenzoylglutamate (pABG) components before re-use, although the lack of a conjugase (gamma-glutamyl hydrolase) would preclude release of pABA itself (Krungkrai et al., 1989). DHPS activity would then be required to reform the 9–10 bond of the folate moiety, and it has been reported that the enzyme from P. berghei can process pABG in addition to pABA in this way, albeit with a Km value ~100-fold less favourable (Ferone, 1973). To characterize the transfectants with respect to folate processing and to assess whether a loss in biosynthetic activity could be compensated for by an increase in salvage, the parasite lines were subjected to radioactive labelling using 3H-folic acid or 3H-folinic acid, and analysed as described above for labelling with pABA. We also wished to investigate how prior reduction of the pteridine ring, as in folinic acid, would influence the nature and levels of the metabolic products observed. The two 3H-labelled compounds were adjusted to the same molar specific activity and cultures were again labelled in parallel together with hypoxanthine uptake controls.

In all three cases, the degree of labelling with folinic acid was significantly higher than with an equal concentration of folic acid of equal specific activity (Fig. 4C and Table 2). Measurements were made from both small (1 ml) and large-scale (40 ml) cultures to assess effects of both scale and the purification/separation of folates from the latter. In either case, both FCB and PKDSM2 were labelled  fourfold more efficiently by folinic acid, and PKDSII about 2.6-fold. However, there was a big difference between FCB on the one hand and the two transfectants on the other as monitored by the ratio of the most highly polyglutamated products to monoglutamates observed when labelled folic acid was used. In the latter pair, this ratio was about 1.7, whereas for FCB, it was only 0.2. With folinic acid, a much greater proportion of the total label was found as polyglutamated products for all three lines, but the same trend was seen (ratios of 5.1 for FCB, 6.8 for PKDSII and 9.6 for PKDSM2). This suggests that in the 24 h period allowed for labelling, the uptake of folic acid and/or the reduction of its fully oxidized pteridine ring imposes a significant kinetic barrier to the eventual production of polyglutamated end-products. However, the greater the degree to which the DHPS activity has been perturbed, the greater is the degree of flux observed through the folate salvage pathway. Thus, both the amount of label incorporated (Fig. 4C) and the proportion of polyglutamated products (Table 2) is highest for PKDSM2, where the residual production of folate via the biosynthetic route is very low, as shown by the labelling with pABA described above.

Growth characteristics of transfectants subject to antifolate drug challenge

Normal parasites are inhibited by the combination of PYR and SDX in a strongly synergistic manner that is essential to their efficacy in antimalarial chemotherapy (Chulay et al., 1984). To assess the role of DHPS in this phenomenon, tests for synergy were carried out on the transfectants together with the host FCB. Such synergy can be illustrated in several ways. The plots of Fig. 5 chart the inhibition of growth by increasing concentrations of each drug alone, of the combination of drugs (PS) in a fixed weight ratio of 10:1 SDX to PYR (comparable to that used in vivo), the arithmetic addition of the effects of the two drugs individually (P+S), and a plot (P×S) which shows the difference between the curve of addition and the actual inhibition exhibited experimentally by the combination. This last illustrates clearly the presence of synergy by taking the form of a hump-shaped curve, with the degree of synergy approximated by the area thereunder. We used this procedure to test the hypothesis that the synergistic effect between the pyrimethamine and sulfadoxine was absolutely dependent upon the presence of DHPS activity. The data clearly showed that where the activity of the enzyme has been reduced to a very low level, as in PKDSM2 (Fig. 5C), the synergistic effect was indeed abolished, whereas in the transfectant carrying the K1-type gene (PKDSII), the synergy was still clearly apparent, though reduced compared with that seen in the untransformed host (Fig. 5A and B). Interestingly, the PKDSM2 line, as well as no longer exhibiting synergy, became much more sensitive to pyrimethamine alone (IC50 value ∼0.5 ng ml−1) than either the host FCB or parasites transfected with a normal dhps gene (IC50 values ∼ 10 ng ml−1 respectively).

Figure 5.

Growth responses of the host (A) and transfectants (B and C) to challenge with sulfadoxine and pyrimethamine. P: pyrimethamine alone; S, sulfadoxine alone; PS: pyrimethamine and sulfadoxine given together in a fixed weight ratio; P+ S: calculated additive effect of pyrimethamine alone plus sulfadoxine alone; P×S: calculated synergistic effect of pyrimethamine with sulfadoxine obtained from the PS and P+S values (see Experimental procedures for detail). The x-axis gives the concentration of SDX in µg ml−1; the weight ratio of SDX to PYR was maintained at 10:1.

Discussion

In common with other organisms, folate metabolism is critically important to the viability of malaria parasites and is targeted in both treatment and prophylaxis of the disease. However, despite a half-century of using antimalarial antifolates, we still lack a detailed understanding of this pathway. To help explore how it might be further exploited and to address unresolved questions of mechanism, we have conducted gene knockout experiments designed to evaluate the importance of folate biosynthesis in the parasite compared to the salvage of preformed folate. Crucially, we have no knowledge of the relative needs of the parasite to use each of these routes, nor of its flexibility in altering the balance between them in adverse circumstances, such as sulfa-drug inhibition of DHPS in the de novo pathway, or depleted exogenous folate levels in the host plasma as a result of nutritional deficiency. Folate is available in the human host plasma principally as 5-methyltetrahydropteroylmonoglutamate (5-MeTHF), but at a very low concentration, averaging around 10 nM (Baker et al., 1994). Moreover, in cases of malnutrition, this can fall to ∼10-fold less than normal (Nelson et al., 2003). Labelling experiments indicate that this form of folate is catabolised by the parasite, with the folate moiety adding to the cofactor pool after removal of the methyl group for use in methionine synthesis (Asawamahasakda and Yuthavong, 1993).

It has been argued that the availability of exogenous folate can reduce or eliminate the importance of DHPS as a target for sulfa-drugs, by providing a bypass to the end products required for essential 1-C transfer reactions (Wang et al., 1997b; Watkins et al., 1997). We thus reasoned that in vitro, transfection mutants with a disrupted dhps gene might be viable if adequate folate supplementation were provided. This was tested using a series of plasmid constructs that varied in the degree to which the wild-type sequence had been altered, to allow for the possibility that complete elimination of DHPS activity might be lethal in this haploid organism, regardless of medium supplementation. Indeed, after transfection with the two 3′-truncated replacement dhps genes, we were never able to detect integrants even after extensive periods of selection, despite the presence of already reduced folate in the form of folinic acid. In contrast, the full-length constructs transfected in parallel under identical conditions reproducibly yielded viable lines with the correct integration within 2 weeks of G418 selection. Formal but unlikely explanations of this difference would be (i) that the 1.29 kb or 1.75 kb of pppk-dhps sequence in the pPKDSneoT1 and pPKDSneoT2 constructs remaining after truncation (compared to a full length of 1.81 kb in pPKDSneoII and pPKDSneoM2) was too short to permit a sufficient frequency of single crossover via homologous recombination for detection, or (ii) that the neomycin phosphotransferase reporter activity is completely abolished by the changes in the DHPS primary sequence adjacent to the fusion point. More probable is that both truncations result in complete disablement of the dhps gene product and that this apparently cannot be tolerated by the parasite. Whereas a complete loss of activity is unsurprising for PKDST1, where over half the DHPS domain would be missing, the results with PKDST2 also indicate that the C-terminal 21 amino acids are indispensable.

Apart from the positive control, the only integrant we could successfully and reproducibly select, PKDSM2, has a full-length dhps domain but is mutated in two active-site residues within the above-mentioned 21 residue region, with the key RVH motif altered to QVQ. This integrant still retained a very low level of activity, as demonstrated by its ability to convert pABA to folate end-products at levels just detectable after HPLC. This was consistent with our measurement of ∼8% residual activity in the equivalent (but unfused) mutant enzyme from T. gondii expressed in E. coli. The difference in viability between PKDSM2 and PKDST2 probably reflects the role of the C-terminal region of the molecule not only in the catalytic step, but also in contributing to the subunit interface in the required dimeric (or higher) structure (Achari et al., 1997; Triglia et al., 1997; Baca et al., 2000). However, although PKDSM2 is viable with folate supplementation, it has lost the ability to survive solely on conversion of pABA as the source of folate in the long term growth test, in contrast with the host FCB and PKDSII.

Importantly, the observation here that a certain level of DHPS activity must be present for parasite viability is in apparent conflict with our earlier studies (Wang et al., 1997b,c), which showed that parasite lines like Dd2 were able to grow almost normally in very high concentrations of SDX if exogenous folate was present, and that even in parasites like HB3, which appear to utilize exogenous folate much less efficiently, growth was never suppressed completely. This suggested that biosynthesis could be dispensed with, if the assumption was made that the high level of SDX (∼25 times the IC50 value measured in folate-free medium) completely inhibits all DHPS activity. The resolution of these seemingly contrary observations might be that DHPS is active in more than one ‘compartment’ in the cell, and that folate salvaged from the medium cannot fulfil the requirements of both (or all) compartments, at least one of which is dependent upon provision of dihydrofolate that is synthesized via the de novo pathway. The corollary of this is that the different compartments would not be equally susceptible to SDX inhibition, with at least one relatively unaffected by the drug. Possible compartments other than the cytoplasm could include the mitochondrion, where folate biosynthesis takes place in a range of organisms, including yeast, plants and mammals (DeSouza et al., 2000; Ravanel et al., 2001), the apicoplast and the nucleus, in the last of which we have preliminary immunofluorescence and EM data indicating the presence of folate pathway enzymes (M. Read, L. Bannister, P. Sims and J. Hyde, unpubl. obs.).

The isotopic labelling and growth studies of the FCB host and dhps-modified transfectants also shed further light on the fate of exogenous folate as it is imported and used by the parasite, and allows us to address a long-standing question about a possible role for DHPS in the salvage pathway. In a previous study using radiolabelled precursors, it was estimated that> 50% of salvaged folate was split into pABG and pterin-6-CHO by the parasite (Krungkrai et al., 1989). Further usage of these components would require DHPS to reform the 9–10 bond of the folate structure by rejoining pABG to the appropriately modified pterin moieties. However, the fact that growth of PKDSM2 with its heavily compromised DHPS was comparable to that of FCB and PKDSII over many cycles when either exogenous folic acid or folinic acid was the predominant source of cofactor suggests that DHPS does not play a major role in salvage. This is supported by our HPLC data, where we saw no evidence of such catabolic intermediates in extracts assayed before purification of folates on the affinity column (data not shown). We suggest therefore that the breakdown of salvaged folate does not in fact occur to an appreciable extent in vivo but that, rather, the folate is used directly after appropriate reduction of the pteridine ring and modification of its 5 position. This would be consistent with the observations using the P. berghei DHPS enzyme that linkage of pABG to the pterin is considerably less efficient than for pABA (Ferone, 1973). Similarly, the E. coli enzyme has a Km for pABA of 2.5 µM with a Ki of 1.3 mM for pABG in competition with pABA (Richey and Brown, 1970). Differences between our metabolic labelling observations and the earlier data (Krungkrai et al., 1989) could possibly stem from the fact that the method used for extracting folates from parasites involved extracting whole blood samples, whereas we purify parasites first. The significance of this is that human erythrocytes contain a latent enzyme activity that can be activated to cleave the 9–10 bond of folates (Braganca et al., 1957; and confirmed in this laboratory). It is possible therefore that the earlier study significantly overestimated the degree of folate breakdown accompanying salvage. However, in agreement with this study, we saw no evidence for the final polyglutamated folate products extending beyond the pentaglutamate level. Indeed, from their elution positions, the major peaks of polyglutamated material from the different parasite lines appeared to correspond more closely to the tetraglutamate standard. Such are the large number of possible intermediates in the reduction of the pteridine ring and its modification at the 5/10 positions for each glutamation level, that the precise nature of these products must await development of a robust mass spectroscopy-based procedure for their separation and analysis.

We can also draw several further conclusions from the metabolic labelling studies. For all of the parasite lines tested, the degree of labelling with folinic acid was c. three to fourfold that obtained using the same concentration of folic acid at the same specific activity. Moreover, in the latter case, a significant proportion of the label was found in the monoglutamate peak, whereas this was only a minor component in the products labelled with folinic acid (Fig. 4). This strongly suggests that the fully oxidized state of folic acid provides a hindrance to either uptake or processing beyond the monoglutamate stage, which significantly reduces the ability of the parasite to efficiently utilize it as a source of salvaged folate. A corollary of this is that the requisite folate reductase activity is likely to be operating at a very low level. Significantly, in both the case of folic and folinic acid, the PKDSM2 mutant was the most highly labelled and the ratio of polyglutamated to monoglutamated products observed was also higher than for the host. This is likely to reflect the fact that this mutant has almost entirely lost its ability to produce folate via the biosynthetic route, as is evident from the data using labelled pABA, and is thus reliant on a much greater proportion of the flux into polyglutamated end-products being carried by the salvage pathway. This is supported by the data for the PKDSII transfectant, which in general are intermediate between those of PKDSM2 and the host, consistent with the fact that, despite having a functional full-length K1-type gene, its DHPS activity is less than that of the host, likely to be due mainly to its juxtaposition with the neophosphotransferase domain.

Comparing all three labels, the patterns observed for pABA in terms of relative amounts of polyglutamated and monoglutamated forms much more closely resembled those of folinic acid rather than folic acid, in that we saw very little residue of monoglutamated material or intermediate forms. At least on this basis, the de novo pathway appears to be a route to final products comparable in efficiency to the salvage pathway, if the latter is provided with already reduced folate, as is the case in vivo (Baker et al., 1994; Nelson et al., 2003). This is supported by quantitative measurements where we adjusted all three labels to identical molar specific activities and corrected cpm values for the difference in 3H and 14C counting efficiencies (data not shown). Similar amounts of total folate product were obtained from both the biosynthetic and salvage routes, although such a comparison does not take into account any pABA synthesized endogenously via the shikimate pathway (Roberts et al., 1998), and thus is not conclusive.

The comparisons of PKDSM2 with PKDSII and host FCB in their responses to the antifolate drugs prove experimentally for the first time that the pronounced synergy between PYR and SDX acting on live parasites is absolutely dependent upon DHPS activity and thus rule out one hypothesis that synergy is due to SDX and PYR acting simultaneously on DHFR (Chulay et al., 1984). This is consistent with studies using purified DHFR from P. falciparum showing negligible synergy at physiological drug levels (Wang et al., 1999). Alternative scenarios have been suggested to account for this synergy, e.g. that the PYR interferes with exogenous folate utilisation and thus considerably increases the reliance of the parasite on the biosynthetic pathway, and thus its susceptibility to sulfa drugs (Sims et al., 1999). Another is that DHPS converts SDX to sulfa-pterin adducts that inhibit activity further downstream, possibly at DHFR as previously speculated (Chulay et al., 1984). Although our current data cannot differentiate between these last two mechanisms, both of which require active DHPS, the PKDSM2 line should provide a useful tool to further investigate this question. Whatever its exact molecular basis, our experiments demonstrated that if DHPS activity is reduced below a critical level, as exhibited by the transfectant PKDSM2, synergy is completely lost. We also observed that such parasites are considerably more sensitive to PYR alone than the host. This could reflect the fact that once parasites have lost most of their ability to make their own folate de novo, pools of dihydrofolate are likely to be diminished and thus compete less effectively with the drug for binding to DHFR, although it does appear as though at least some of this loss is offset by increased flux through the salvage pathway. A precedent for this view is a recent study (Nzila et al., 2003) showing that where such pools are reduced by inhibiting the influx of folate with the drug probenecid, sensitivity to PYR is also markedly increased.

More generally, the haploidy of P. falciparum throughout most of its life cycle complicates knockout studies of this type where genes likely to be essential are involved. Although this may eventually be overcome by the use of an inducible promoter system that ideally would allow downregulation of the activity of interest to a desired level, our current approach of reducing the activity by targeted mutagenesis provides a viable alternative and has the advantage that the test gene is still expressed from its unaltered endogenous promoter, thus permitting more valid comparisons with the unmodified host parasite line.

Experimental procedures

Materials

p-amino[ring-14C]benzoic acid (55 mCi mmol−1, 0.1 mCi ml−1) and [3′, 5′, 7, 9-3H]leucovorin (folinic acid; 42 Ci mmol−1, 1 mCi ml−1) were purchased from Moravek Biochemicals, CA. [3′, 5′, 7, 9-3H]folic acid (24 Ci mmol−1, 1 mCi ml−1) and [8-3H]hypoxanthine (28 Ci mmol−1, 1 mCi ml−1) were from Amersham, UK. Pteroyldi-gamma-glutamic acid, pteroyltri-gamma-glutamic acid, pteroyltetra-gamma-glutamic acid and pteroylpenta-gamma-glutamic acid were from Schircks Laboratories, Switzerland. Folic acid and folinic acid were from Sigma, UK.

Parasite lines and transfection

The P. falciparum line FCB was grown and transfected as previously described (Wang et al., 2002) with slight modification. Briefly, parasites were routinely cultured in RPMI 1640 (Invitrogen), 2.5% O+ human red cells, supplemented with 0.5% Albumax II (Invitrogen) or albumin fraction V (A1933, Sigma), 20 µg ml−1 gentamycin, 0.2% glucose, 1 µg ml−1 folinic acid and 5 µg ml−1 of hypoxanthine (Sigma). Supplementation with folinic acid (over and above the folic acid already present in standard RPMI 1640) was found to increase the health of the electroporated cultures and accelerate the appearance of transfectants from about 3 weeks post-G418 selection to about 2 weeks. The transfection was achieved by inoculating red cells preloaded with the desired plasmid construct (Deitsch et al., 2001), using a Bio-Rad gene pulser. The following plasmids were designed for use in the bsd/neo positive selection system (Wang et al., 2002): pPKDSneoII and pPKDS are described in Wang et al. (2002); PKDSneoM2 was derived from pPKDSneoII by site-directed mutagenesis, using Pwo DNA polymerase (Roche) and complementary primers 5′-A GGT AGATCTA ATAcaAGTTCAaGACGTTTTAGAAACAA AATCGG and 5′-CCGATTTTGTTTCTAAAACGTCtTGAACT tgTATTAGATCTACCT, where lower case letters indicate changes from the wild-type sequence (acc. no. Z30655). These convert the conserved RVH amino-acid motif in the DHPS domain to QVQ. Plasmids pPKDSneoT1 and pPKDSneoT2 were constructed as described (Wang et al., 2002), but using primers C498 (5′-GATTTCCTCTAGAATG CATTAGAACTA) and C600 (5′-TAAAACGTCATGACTAGT TATTAGATCTAC), respectively, to truncate the dhps domain at different positions (see Table 1). Primers for the neo cassette (Wang et al., 2002) were modified for compatibility  with the restriction sites underlined (XbaI and SpeI).

PCR, RT-PCR and Southern blotting to monitor the transfection process and transcription from the integrants were carried out as described (Wang et al., 2002). DNA blots were probed with a PCR fragment generated with primers C481 and C555 (Wang et al., 2002) from the pppk-dhps gene of K1 and labelled with a DIG kit (Roche). RNA was isolated using an RNeasy miniprep kit (Qiagen) and treated with DNase I to eliminate any contamination from genomic DNA.

Folic acid, folinic acid and pABA requirements of host and transfectants

Parallel cultures were initiated in custom RPMI 1640 (Invitrogen) devoid of pABA and folic acid, to which either pABA, folic acid or folinic acid had been added back (1 µg ml−1), sampled every 48 h and parasite numbers per microlitre culture recorded. The cultures were subcultured when parasitaemias approached 5–7% by dilution to a parasitaemia of ≤ 1%, and the degree of dilution recorded. The parasite population number per microlitre was calculated by multiplying the observed number per microlitre by the dilution factors recorded during the period of culture.

Metabolic radiolabelling of parasite lines

Plasmodium falciparum was cultured as above in 40 ml RPMI 1640 (4–5% haematocrit) and synchronized with sorbitol (Read et al., 1993). Just before labelling, cells were washed twice with PBS and resuspended in pABA- and folate-depleted RPMI 1640. Labelling was initiated when the parasites were in the early ring stage and at a concentration of 1.3–2.0 × 104µl−1 of culture. The radioactive metabolic precursors p-amino[14C]benzoic acid, [3′, 5′, 7, 9-3H]folic acid and [3′, 5′, 7, 9-3H]folinic acid) were added individually and the culture continued for 24 h at 37°C before harvesting. Cells were lysed with 200 µl of 10% saponin (0.05% final concentration) and the parasites pelleted at 2500 g for 5 min. To monitor the growth of the cultures, a 1 ml aliquot was initially removed, mixed with 0.5 µCi 3H-hypoxanthine and cultured in parallel for the same period. This culture was harvested at the same time as the main culture and the hypoxanthine incorporation measured as described (Wang et al., 1997c). Small-scale cultures (1 ml) were also used to compare folate/pABA labelling efficiencies. These were processed as above, except that purification on the folate binding protein column and HPLC separation of products were omitted.

Extraction of folates from parasites and affinity purification

Parasite pellets were washed three times in 1 ml PBS to remove any extracellular label and resuspended in 1 ml extraction buffer (0.1 M Tris-HCl, 1% ascorbic acid, 0.8% NaCl, pH 7.0), boiled for 10 min and centrifuged at 10 000 g for 10 min. The supernatants were stored at − 20°C until HPLC analysis. Folate binding columns were made by covalent coupling of 1 mg folate binding protein (Sigma) to a 1 ml HiTrap NHS-activated HP column (Amersham Biosciences). The column was washed with 4 ml of 0.02 M trichloroacetic acid (TCA), 2 ml of water and then equilibrated with 3 ml 1 M sodium phosphate buffer, pH 7. Extracted folate (1 ml) was loaded onto the column, which was then washed with 3 ml 1 M sodium phosphate buffer, pH 7.0 and 1 ml water. Bound folate was eluted with 2 ml 0.02 M TCA. Aliquots (10 µl) from the pre- and postcolumn solutions were counted on a liquid scintillation counter to monitor the yields from the column.

HPLC analysis of folate derivatives

High performance liquid chromatography was performed on a Dionex Summit liquid chromatography system. Radioactivity was measured by an EG and G Berthold detector with a dry cell connected online to the HPLC system in series with the UV detector. The protocol for folate analysis was adapted from Selhub (1989) and carried out on a C18 Econosphere (5 µm, 150 × 4.6 mm) column with a C18 Econosphere guard column attached. Folate standards (10 µg per compound) were added to test samples before injection and the mixture eluted with buffer gradient at a flow rate of 1.0 ml min−1. The gradient was formed from two solutions each containing 5 mM tetrabutylammonium phosphate (TBAP), 25 mM NaCl, 10 mM Tris-HCl, pH 6.8. Buffer A was made in water, Buffer B was made in 65% acetonitrile and 35% water. The column was equilibrated with a mixture containing 90% of A and 10% of B. After sample injection, the column was washed for 2 min with the same mixture. Thereafter the proportion of B in the elution was increased with time to reach 30% at 5 min, maintained at 30% until 9 min, then raised to 45% at 15 min, maintained at this concentration until 25 min, then raised to 100% at 30 min. The eluent was monitored at 280 nm and by the on-line radiodetector, the retention times recorded and peak areas integrated. Eluted materials were also collected in 380 µl fractions for further analysis of the peaks and for more sensitive counting in a Wallac 1409 DSA liquid scintillation counter.

Assay of synergy between pyrimethamine and sulfadoxine

The synergy test was adapted from the sulfadoxine susceptibility assay described earlier (Wang et al., 1997c) with slight modifications. For more consistent growth, the parasites were tested in medium containing 25 ng ml−1 of folic acid, 25 ng ml−1 of pABA instead of folate/pABA-free medium. Both SDX and PYR were serially diluted in steps of 2× in DMSO. To calculate the degree of synergy, three separate tests are required. The first assays the response of the parasites to SDX, the second to PYR and the third to SDX and PYR combined at the same concentrations over the same series of dilutions as in tests one and two. One µl of the diluted drug was applied to each well in a 96-well microtitre tray. Then 200 µl of parasite culture were added to each well at a parasitaemia of 0.5%, the tray sealed in a container under 5% CO2, 5% O2, 90% N2 and incubated at 37°C for 48 h before labelling by addition of 0.5 µCi 3H-hypoxanthine in 25 µl of culture medium. After incubation for 18 h the cultures were harvested and the uptake of the 3H-hypoxanthine measured in the scintillation counter. The inhibition value (V) for each well is calculated by the following general formula:

VI = (CPM0 – CPMI)/CPM0) × 100, where CPM0 is the hypoxanthine uptake value of the control well without inhibitor and CPMI is the hypoxanthine uptake value of the well with added inhibitor (either SDX or PYR or both). The additive effect of SDX (A) and PYR (B) is calculated from: Vadd = [1 – (CPMA × CPMB)/(CPMA0 × CPMB0)] × (100). The synergy of SDX and PYR is calculated by Vsyn = VAB – Vadd; CPMA, CPMB, CPMAB are the hypoxanthine uptake values of the cultures in the well with inhibitor A, B, and A plus B added, respectively; CPMA0, CPMB0 are the corresponding hypoxanthine uptake values of the control cultures without inhibitors A and B respectively.

Acknowledgements

We thank the Wellcome Trust, UK (grant no. 056845) for financial support.

Ancillary