Introduction

Top Summary Introduction Materials & Methods Results Discussion References

The rapid spread of Plasmodium falciparum strains that are resistant to chloroquine and pyrimethamine/sulfadoxine (Fansidar) underscores the need for pharmacological initiatives to counter the resulting increases in malaria mortality and morbidity rates. New antimalarial agents in such initiatives may be derived as novel compounds or modifications of existing drugs. One prophylactic drug that has undergone a resurgence of interest for use against malaria is the biguanide proguanil (Paludrine); this compound is cyclized by hepatic cytochrome P450 isoenzymes to the active metabolite cycloguanil, which has been reported to act on P. falciparum DHFR (EC 1.5.1.3). This enzyme, which in P. falciparum is fused with TS as a homodimeric bifunctional protein, catalyzes the NADPH-dependent reduction of dihydrofolate to tetrahydrofolate, thus providing a source of one-carbon donors for methyl transfer reactions and dTMP synthesis required for parasite survival.

Evidence that cycloguanil acts on P. falciparum DHFR comes from the findings of associations between the activity of this metabolite and parasite DHFR point mutations, as well as from more recent inhibition and binding studies with P. falciparum DHFR variants expressed in Escherichia coli (Foote et al., 1990; Peterson et al., 1990; Sirawaraporn et al., 1997). The possibility that this metabolite may also have activity against a second target has been raised by reports that levels of susceptibility to cycloguanil could vary up to 8-fold among parasite isolates encoding identical DHFR sequences (Foote et al., 1990; Basco et al., 1995). The presence of a secondary target has also been proposed to explain the finding that cycloguanil-induced depletion of dTTP pools was not readily reversed by folinic acid (Yeo et al., 1997). Recently, we found that intraparasitic expression of MTX-resistant human DHFR (which is innately resistant to antimalarial agents and can overcome the metabolic block resulting from inhibition of the parasite DHFR enzyme) resulted in a 10-fold increase in resistance to cycloguanil, as opposed to the >1000-fold increase in resistance to the DHFR inhibitors MTX and WR99210 (Fidock and Wellems, 1997). This discrepancy could have resulted either through the action of cycloguanil on a secondary target whose inhibition could not be complemented by human DHFR or because these assays were performed on a parasite line (FCB) harboring the DHFR mutations Val16 and Thr108, a pair of mutations that render the parasite enzyme resistant to cycloguanil and thus might minimize the protective effect of the resistant human enzyme.

Here we report molecular complementation assays that address the role of parasite DHFR in relation to cycloguanil and its parent compound. These studies use a new system for transformation, in which P. falciparum parasites expressing either the wild-type or the MTX-resistant L22Y form of human DHFR are selected with the dihydrotriazine WR99210. Results clearly distinguish separate modes of pharmacological action for proguanil and cycloguanil, demonstrate that the only significant action of cycloguanil is against parasite DHFR, and may suggest important avenues of investigation into classes of lead compounds that could be used in combination against these targets.

	Materials and Methods

Top Summary Introduction Materials & Methods Results Discussion References

Plasmid constructs. Wild-type, human, full-length dhfr cDNA was amplified from the pDS5+wt construct (a kind gift from Raymond Blakley, St. Jude's Hospital, Memphis, TN) using the primers 5'-cctttttatgcatggttcgctaaactgcatcg and 5'-aatttcaagcttaatcattcttctcatatacttc. After cloning into the pCR2.1 vector (Invitrogen, Carlsbad, CA), this 0.6-kilobase gene was inserted as a NsiI/HindIII fragment in the place of the luciferase gene in the pHLH-1 construct (Wu et al., 1995), yielding pHDWT. The dhfr sequences in both pHDWT and the previously reported construct pHD22Y (Fidock and Wellems, 1997) contain silent mutations at nucleotide positions 90 and 96 (codons 29 and 31; Genbank accession number V00507), which result in loss of an EcoRI restriction site. For transfections, plasmid DNA was purified over two cesium gradients or Qiagen Maxiprep columns (Chatsworth, CA) and concentrated on a Centricon-100 column (Amicon, Beverly, MA) in incomplete Cytomix (120 mM KCl, 0.15 mM CaCl₂, 2 mM EGTA, 5 mM MgCl₂, 10 mM K₂HPO₄/KH₂PO₄, 25 mM HEPES, pH 7.6).

Parasites. P. falciparum, asexual, blood-stage parasites, were propagated in leukocyte-free human RBC (at 4% hematocrit) in complete medium [RPMI 1640 with L-glutamine (catalog no. 31800; Life Technologies, Gaithersburg, MD), 50 mg/liter hypoxanthine, 10 mg/liter gentamycin, 25 mM HEPES, 0.225% NaHCO₃, and 0.5% Albumax I (Life Technologies)] and were grown at 37° in tissue culture flasks gassed with 5% CO₂/5% O₂/90% N₂. Parasites were transfected using electroporator settings of 0.31 kV and 960 µF. These modified settings result in increased efficiency of transfection, relative to previous high-voltage/low-capacitance settings (Wu et al., 1995), as measured using transient luciferase activity assays (Fidock and Wellems, 1997). Although these newer low-voltage/high-capacitance settings result in greater initial RBC lysis, they consistently result in a 1-2-day reduction in the number of days required to microscopically observe transfected parasites.

Drug assays. MTX (Sigma Chemical, St. Louis, MO), WR99210 (a kind gift from David Jacobus, Jacobus Pharmaceuticals, Princeton, NJ), and cycloguanil and proguanil (kind gifts from Dennis Kyle and Wilbur Milhous, Walter Reed Army Institute of Research, Washington, DC) were maintained at 80° as 10 mg/ml stock solutions in dimethylsulfoxide (Sigma). The structures of these drugs are shown in Fig. 1. For the drug assays, serial 2-fold drug dilutions were made in complete medium modified to contain 2.5 mg/liter hypoxanthine ("low-hypoxanthine medium"). These dilutions were added to 96-well culture plates at 100 µl/well. Parasites were diluted to a 2-fold concentrated stock solution consisting of 0.5-1.0% parasitemia and 4% hematocrit in low-hypoxanthine medium and were added at 100 µl/well. After 48 hr of incubation, 100 µl of culture supernatants were replaced with 100 µl of low-hypoxanthine medium containing [³H]hypoxanthine at a concentration of 7.5 µCi/ml. After an additional 24 hr, supernatants were removed, distilled water was added (200 µl/well), and the plates were frozen and thawed before cells were harvested onto glass fiber filters (Wallac, Turku, Finland). Air-dried filters were placed in sample bags (Wallac) and immersed in scintillation fluid (Ecoscint A; National Diagnostic, Atlanta, GA), and radioactive emissions were counted in a 1205 Betaplate reader (Wallac). Mean emission levels were generally in the range of 20,000-30,000 cpm. Percentage reduction in hypoxanthine uptake (a marker of growth inhibition) was calculated as follows: reduction = 100 × [(geometric mean cpm of no-drug samples)  (mean cpm of test samples)]/(geometric mean cpm of no-drug samples). These percentage reductions were used to plot data as a function of drug concentrations. IC₅₀ values were determined by linear regression analyses of the linear segments of the curves.

View larger version (18K): [in this window] [in a new window]   Fig. 1.   Structures of agents active against P. falciparum and central to this study. Cycloguanil and WR99210 are the active metabolites of the biguanides proguanil and PS-15, respectively.

	Results

Top Summary Introduction Materials & Methods Results Discussion References

WR99210 selection of P. falciparum parasites transformed with human wild-type or MTX-resistant dhfr genes. We transformed the P. falciparum lines 3D7 (cycloguanil-sensitive) and FCB (cycloguanil-resistant) with plasmids expressing wild-type and MTX-resistant human DHFR, respectively. Plasmids pHDWT and pHD22Y differ only at codon 22 of the human dhfr gene, with the former encoding a wild-type leucine and the latter encoding a tyrosine residue responsible for MTX resistance. Expression of human DHFR in these plasmids is under the control of P. falciparum regulatory elements, namely the 5'-untranslated region of the hrp3 gene and the 3'-untranslated region of hrp2 (Fig. 2).

View larger version (30K): [in this window] [in a new window]   Fig. 2.   Map of the human dhfr vectors pHDWT and pHD22Y used for transfection of P. falciparum. The L22Y mutation (*) that distinguishes these two vectors is located within the folate binding pocket and results in a >3000-fold decrease in MTX affinity in the presence of dihydrofolate and NADPH, while leading to only minimal decreases in catalytic efficiency at physiological pH (Lewis et al., 1995).

Antifolate responses of human dhfr-transformed parasites. Parasites episomally transformed with either pHDWT (3D7/pHDWT) or pHD22Y (FCB/pHD22Y) were assayed for their susceptibility to WR99210 and MTX. As controls, we included nontransformed parasites (3D7 and FCB) and also parasites that had been previously derived from the episomally transformed lines and propagated in the absence of drug for 15 generations. This regimen is known to lead to the rapid outgrowth of parasites that have lost episomes (Wu et al., 1996). Plasmid rescue from these cultures (3D7/no pHDWT and FCB/no pHD22Y) revealed >99% reduction in episome numbers, and polymerase chain reaction assays showed no detectable integration of the vector into the genome (data not shown).

View larger version (22K): [in this window] [in a new window]   Fig. 3.   Inhibition levels of P. falciparum lines transformed with human wild-type or L22Y dhfr plasmids (pHDWT or pHD22Y), measured as the percentage reduction of hypoxanthine uptake as a function of WR99210 (A) and MTX (B) concentrations, shown on a logarithmic scale (log4). The episomally transformed lines 3D7/pHDWT and FCB/pHD22Y were maintained under continuous drug pressure for 2 months after DNA electroporation. Nontransformed parasites and lines cured of episomes (3D7/no pHDWT and FCB/no pHD22Y) were included as controls. Results presented in Figs. 3-5 were obtained in single representative experiments. In repeated experiments, variations of up to 30% were observed in absolute IC50 values, without affecting the relative values for the different parasite lines (see Materials and Methods). At the single WR99210 drug concentration of 0.16 nM, the nontransformed and cured 3D7 lines showed variation that was not observed in repeated experiments or in the response to MTX.

View larger version (25K): [in this window] [in a new window]   Fig. 4.   Inhibition levels of parasites expressing human DHFR from genome-integrated, single-copy genes or episomally based, multicopy genes, as a function of WR99210 (A) and MTX (B) concentrations, shown on a logarithmic scale (log4). The clone FCB/pHD22Y/integrant/A5-C8 expresses a single copy of the L22Y human dhfr gene from a chromosomal location, after recombination and integration of pHD22Y into the nuclear genome. Polymerase chain reaction and Southern hybridization assays demonstrated the absence of episomally replicating plasmids in this clone. Drug responses were tested in parallel in the independently generated control lines FCB/pHD22Y/MTX/episomal and FCB/pHD22Y/WR99210/episomal, which had been transfected with the pHD22Y vector 1 month earlier and selected using MTX and WR99210, respectively. These cultures were confirmed to be episomally replicating the human dhfr construct, as shown by plasmid rescue and Southern hybridization assays (data not shown).

Cycloguanil, but not proguanil, targets P. falciparum DHFR. We tested the effect of the metabolite cycloguanil and its parent compound proguanil on human DHFR-transformed 3D7 parasites. Results showed that, whereas 3D7 and the control line 3D7/no pHDWT (transformed but subsequently cured of its episomes) were sensitive to cycloguanil, with IC₅₀ values of 2 nM, the IC₅₀ of the human dhfr-transformed 3D7 line was 7 µM (an increase of 3500-fold) (Fig. 5A). This indicates a very high degree of protection afforded by the human enzyme against the metabolic target of this drug; therefore, by inference, cycloguanil specifically targets P. falciparum DHFR to the exclusion of any other major target. Parallel tests on FCB parasites carrying the Val16 and Thr108 DHFR mutations, which correlate with cycloguanil resistance, revealed an IC₅₀ of 0.5-0.6 µM in nontransformed controls, compared with an IC₅₀ of 2.6 µM in human dhfr-transformed FCB parasites. This indicates that the FCB mutant DHFR enzyme is almost as resistant to cycloguanil as its human counterpart.

View larger version (25K): [in this window] [in a new window]   Fig. 5.   Inhibition levels of cycloguanil-sensitive (3D7) versus -resistant (FCB) P. falciparum lines episomally transformed with human dhfr, as a function of cycloguanil (A) and proguanil (B) concentrations, shown on a logarithmic scale (log4 and log2, respectively). Control lines are the same as those described in the legend to Fig. 3.

	Discussion

Top Summary Introduction Materials & Methods Results Discussion References

Using complementation assays based on expression of human DHFR in drug-sensitive malaria parasites, we show that the only significant antimalarial target of cycloguanil is parasite DHFR. These assays indicate a separate mode of action for the parent compound proguanil. Evidence in favor of a pharmacologically important role for the parent compound comes from early findings that it was considerably more effective than its cycloguanil metabolite in treating malaria in humans or monkeys, with subsequent work demonstrating that this was not attributable to pharmacokinetic differences (Schmidt et al., 1952; Robertson, 1957; Smith et al., 1961). This is confirmed by reports that the use of proguanil for antimalarial prophylaxis or treatment can be equally effective in individuals who do not metabolize proguanil to cycloguanil (Ward et al., 1989; Mutabingwa et al., 1993; Mberu et al., 1995).

From these studies, we propose that proguanil acts as an antimalarial agent in its native form as well as in the form of its active metabolite cycloguanil. This dual activity helps to explain why the proguanil/atovaquone (Malarone) combination provides almost 100% efficacy in treating P. falciparum malaria in areas in which cycloguanil resistance can be high and administration of atovaquone alone results in very rapid selection of resistant parasites (Blanchard et al., 1994; Looareesuwan et al., 1996; Radloff et al., 1996; Lell et al., 1998). Such a "triple-drug" effect of proguanil/atovaquone may account for its usefulness against malaria in regions that already show a high prevalence of drug-resistant strains. Furthermore, the structure and dual activities of proguanil might serve as a valuable starting point for the development of new antimalarial agents that are effective as both therapeutic and prophylactic agents.

The WR99210/human dhfr transformation system described here has several advantages over previous methods of genetically modifying P. falciparum parasites, which have relied on the use of pyrimethamine or MTX to select parasites expressing pyrimethamine-resistant DHFR enzymes of protozoan origin or human MTX-resistant DHFR (Crabb and Cowman, 1996; Wu et al., 1996; Fidock and Wellems, 1997). The pyrimethamine system is restricted, in that it cannot be used in selection schemes with parasites from field isolates and laboratory lines that are already resistant to this drug. MTX is highly toxic to mammalian cells (Huennekens, 1994) and can select nontransformed MTX-resistant P. falciparum mutants, even at high drug concentrations. In contrast, WR99210 remains effective against parasites having a range of DHFR point mutations associated with high-level resistance to pyrimethamine and/or cycloguanil. Furthermore, this drug has not selected resistant mutants in 15 independent transformation experiments (using the lines 3D7, FCB, HB3, and Dd2, which differ at the dhfr locus). WR99210 also shows potential for in vivo selection studies. Indeed, we recently used this agent to select human dhfr-transformed rodent P. berghei parasites in vivo (de Koning-Ward T and Fidock D, unpublished observations). More generally, the proven efficacy of WR99210 or its prodrug PS-15 against the opportunistic pathogens T. gondii, Pneumocystis carinii, and Mycobacterium avium complex (Hughes et al., 1993; Meyer et al., 1995; Brun-Pascaud et al., 1996) raises the possibility that this WR99210/human dhfr system might be broadly applicable to genetic manipulation of infectious microorganisms that are susceptible to this agent.

The transformation system described here provides a new approach for screening antimalarial compounds. Assays using this system could be used to identify lead DHFR inhibitors, whose specificity for parasite DHFR would be indicated by the relative drug activities against nontransformed versus human dhfr-transformed parasite lines. This would provide valuable preliminary information on the therapeutic value of these compounds. Use of a set of P. falciparum lines encompassing the known range of dhfr variants would also allow measurements of activity against drug-resistant strains. These assays have the advantage of directly testing compounds against parasites expressing native DFHR and enable comparisons of activities against parasite and human DHFR enzymes in the same host cell. This provides an alternative to systems that promote screening using recombinant DHFR enzymes expressed in E. coli or P. falciparum dhfr-transformed yeast (Brobey et al., 1996; Wooden et al., 1997).

For combination chemotherapy, we propose that one promising strategy is to combine multiple parasite-specific DHFR inhibitors that interact with different residues surrounding the active site, to form multiple hits on the same target enzyme. This approach would complement traditional combination chemotherapeutic strategies that advocate the use of two or more compounds that act on separate targets to produce additive or (preferably) synergistic effects. The rationale for combining multiple compounds that act against P. falciparum DHFR comes from the findings that different patterns of resistance to individual DHFR inhibitors arise from separate sets of mutations affecting the folate substrate pocket (Cowman et al., 1988; Peterson et al., 1988, 1990; Foote et al., 1990; Sirawaraporn et al., 1993, 1997). As a consequence, certain combinations of DHFR inhibitors (for example, WR99210 and cycloguanil) display little or no cross-resistance (Toyoda et al., 1997; Wooden et al., 1997). Furthermore, it is known that particular combinations of drug-resistant point mutations can totally ablate parasite DHFR enzyme function in vitro (Sirawaraporn et al., 1997; Reynolds and Roos, 1998), suggesting that these combinations would be excluded from natural parasite populations. Thus, the combination of multiple DHFR inhibitors such as cycloguanil and WR99210 (or improved derivatives) might be an effective strategy to counter the spread of drug-resistant P. falciparum by preventing the selection of multidrug-resistant DHFR enzymes. This approach might be particularly effective in Africa, where cycloguanil-resistant dhfr genotypes have not been found in extensive molecular epidemiological investigations (Basco et al., 1995; Plowe et al., 1997; Wang et al., 1997). Such a strategy could be incorporated into a combination chemotherapeutic approach that attacks multiple targets, including the target of proguanil and DHFR or other enzymes involved in the folate biosynthetic pathway, in a manner analogous to that of triple- or quadruple-drug combinations now being used to treat human immunodeficiency virus or tuberculosis infection (Lipsky, 1996; Reichman, 1996). This calls for a change in thinking about the treatment of malaria in regions harboring multidrug-resistant P. falciparum strains.