Inhibition of p-Aminobenzoate and Folate Syntheses in Plants and Apicomplexan Parasites by Natural Product Rubreserine*

Background: pABA biosynthesis is a potential target for antifolate drugs. Results: Rubreserine inhibits GAT-ADCS, an enzyme involved in pABA biosynthesis, and decreases the folate content in Arabidopsis and Toxoplasma. Conclusion: Specific inhibition of pABA synthesis induces growth limitation of plants and apicomplexan parasites. Significance: GAT-ADCS is a valuable target in eukaryotes, and rubreserine is a novel scaffold for anti-parasitic drugs. Glutamine amidotransferase/aminodeoxychorismate synthase (GAT-ADCS) is a bifunctional enzyme involved in the synthesis of p-aminobenzoate, a central component part of folate cofactors. GAT-ADCS is found in eukaryotic organisms autonomous for folate biosynthesis, such as plants or parasites of the phylum Apicomplexa. Based on an automated screening to search for new inhibitors of folate biosynthesis, we found that rubreserine was able to inhibit the glutamine amidotransferase activity of the plant GAT-ADCS with an apparent IC50 of about 8 μm. The growth rates of Arabidopsis thaliana, Toxoplasma gondii, and Plasmodium falciparum were inhibited by rubreserine with respective IC50 values of 65, 20, and 1 μm. The correlation between folate biosynthesis and growth inhibition was studied with Arabidopsis and Toxoplasma. In both organisms, the folate content was decreased by 40–50% in the presence of rubreserine. In both organisms, the addition of p-aminobenzoate or 5-formyltetrahydrofolate in the external medium restored the growth for inhibitor concentrations up to the IC50 value, indicating that, within this range of concentrations, rubreserine was specific for folate biosynthesis. Rubreserine appeared to be more efficient than sulfonamides, antifolate drugs known to inhibit the invasion and proliferation of T. gondii in human fibroblasts. Altogether, these results validate the use of the bifunctional GAT-ADCS as an efficient drug target in eukaryotic cells and indicate that the chemical structure of rubreserine presents interesting anti-parasitic (toxoplasmosis, malaria) potential.

commonly used in clinical treatments against parasites of the Apicomplexa phylum, such as Plasmodium falciparum or Toxoplasma gondii (8,9). However, the use of these drugs is compromised by the emergence of resistance, and the currently used chemical scaffolds and protein targets are actually overexploited. Nevertheless, the long established efficacy of folate metabolism as a clinical target is strongly encouraging to identify new inhibitors acting against other enzymes of the folate pathway comprising nine enzyme activities in addition to DHPS and DHFR (Fig. 1A) (10 -12).
Among potential targets, the enzymes involved in the pABA branch of the pathway are of great interest (13). Indeed, they are absent in animals, and the only known metabolic fate of pABA is its commitment in folate synthesis. In addition, it was recently shown in plants that the production of pABA, together with the production of pterins, is rate-limiting for the whole folate pathway (4,5). Also, it was shown in P. falciparum that pABA was an effective salvage substrate in experiments using antifolates, suggesting that pABA metabolism might offer opportunities for chemotherapy (14). The pABA moiety is synthesized in two steps from chorismate (a metabolite also involved in aromatic amino acid synthesis (15)). First, chorismate is aminated to form 4-amino-4-deoxychorismate (ADC), and ADC is then aromatized with loss of pyruvate (Fig. 1B). In many bacteria such as Escherichia coli or Bacillus subtilis, ADC synthesis requires two separate proteins, PabA (a glutamine amidotransferase) and PabB (the ADC synthase) (16). In eukaryotes, the situation appears different. Indeed, in plants and lower eukaryotes, such as yeast and Apicomplexa, ADC synthesis is catalyzed by a single bifunctional protein (Fig. 1A) containing two domains, the glutamine amidotransferase (GAT) in the N-terminal part and the ADC synthase (ADCS) in the C terminus (17). Based on sequence similarities with TrpG (the component II of anthranilate synthase), GAT-ADCS is classified as a member of the G-type group of amidotransferases (18,19). There is only one gene coding for GAT-ADCS in apicomplexan parasites and plants, and a mutation in the plant gene is embryo-defective. The ADCS domain belongs to the group of chorismate-utilizing enzymes, which also contains salicylate synthase, isochorismate synthase, and anthranilate synthase (20,21). Until now, searches of inhibitors for this class of enzymes were only achieved using prokaryotic systems. They involved docking studies and design of chorismate analogous compounds (22)(23)(24)(25), combinatorial chemistry approaches (26,27), and a specific screening of a microorganism extract collection using growth inhibition of test bacteria as a marker of activity (28,29). Several compounds were identified by these different methods but appeared to be relatively weak inhibitors of ADCS, although some of them could be quite potent against other chorismate-utilizing enzymes (22)(23)(24)26). The most potent inhibitor of ADCS reported to date is an analog of chorismate (2-hydroxy-4-amino-4-deoxychorismate), exhibiting a K i value of 38 M against the purified enzyme (23). To our knowledge, the in vivo effects of these ADCS inhibitors have not been investigated.
In this report, using a purified recombinant plant GAT-ADCS as a model enzyme for bifunctional GAT-ADCS, we screened a chemical library for new inhibitors of pABA synthesis. We identified one compound exhibiting a K i Ͻ 10 M and measured the impact of this molecule on a plant (Arabidopsis thaliana) and two apicomplexan parasites (T. gondii and P. falciparum).

EXPERIMENTAL PROCEDURES
Materials-A. thaliana (ecotype Columbia) seedlings were grown on plates containing Murashige and Skoog medium, 15% agar, plus the various molecules to be tested. Seeds were first sterilized by soaking for 15 min in a solution containing 0.095% Tween and 0.57% sodium hypochlorite before placing on the agar medium. The plates were conserved in the dark at 4°C for 48 h and then transferred in a greenhouse (20°C, 80% humidity, 150 moles photon m Ϫ2 ⅐s Ϫ1 , 12 h light period). The number of  F-Glu n , the polyglutamylated form of H 4 F. Merged symbols represent bifunctional enzymes. The activities 9 and 11 are catalyzed by a single enzyme in E. coli and Plasmodium. In Plasmodium, the activity 6 remains to be identified. Above the broken line, the activities are absent in animals. The black stars show the two enzymes actually targeted by antifolate drugs. The white stars show the potential target studied in this work. B, details of the two-step pathway required for pABA synthesis. In all eukaryotes, the first step involves a bifunctional GAT-ADCS, whereas two separate proteins PabA and PabB are found in most (but not all) prokaryotes (19). seedlings at the two leaf stage (rosette stage) was counted after 2 weeks.
A. thaliana (ecotype Columbia) cell suspension cultures were grown and subcultured as described previously (30). For measurements of metabolites, cells were collected after 7 days of treatment, rapidly washed with distilled water, weighted, frozen in liquid nitrogen, and stored at Ϫ80°C for later analyses.
T. gondii tachyzoites from the RH-YFP 2 strain (kindly provided by B. Striepen, Athens, GA) were propagated in human foreskin fibroblasts (HFF) under standard procedures as described previously (31). Invasion and proliferation assays were performed on HFF cells grown to confluence on glass coverslips in 4-or 24-multiwell plates. For the invasion assay, freshly egressed RH-YFP 2 parasites were incubated for 5 h with or without the different drugs. To perform synchronized invasion, 10 6 parasites/well were centrifuged for 30 s at 1300 rpm onto HFF monolayers, and wells were incubated for 15 min in a water bath at 37°C. Wells were further washed three times with cold PBS to eliminate extracellular parasites. Infected cells were fixed in 5% formaldehyde/PBS for 30 min and stored in PBS at 4°C until staining. To distinguish intracellular from remaining extracellular parasites, coverslips were incubated with the primary antibody mAb Tg05-54 against the major Toxoplasma surface protein SAG1 (TgSAG1) and then with Texas Red-conjugated goat anti-mouse secondary antibody (Molecular Probes). Intracellular parasites exhibiting a faint red color can easily be distinguished from extracellular parasites that are bright red. Nuclei were stained with Hoechst 33258 (Molecular Probes). The number of intracellular parasites was determined from 12 randomly selected fields per coverslip and per experiment with a Zeiss Axioplan 2 microscope equipped for epifluorescence and phase contrast. Invasion was expressed as percent of the number of intracellular parasites recorded in nontreated cells.
To assess the effect of the drugs on intracellular growth of T. gondii (proliferation assay), HFF monolayers were infected with 10 5 parasites/well (cf. invasion assay). Wells were then washed three times with PBS to eliminate extracellular parasites, and drugs were added. After 24 h at 37°C in a humidified atmosphere containing 5% CO 2 , cells were fixed and stained with Hoechst 33258 as described above. For each drug concentration, the number of parasites was determined from at least 100 individual vacuoles.
To determine folate concentrations, parasites were grown for 24 h in HFF cells in the presence or absence of 20 or 40 M rubreserine. The infected HFF monolayers were washed three times with PBS, harvested with a cell scraper, and passed four times through a 27-gauge needle to break the cells. Broken cells and released parasites were washed three times with PBS, and then parasites were purified by filtration through a 3 m Nucleopore membrane, further washed two times in PBS, concentrated by centrifugation, and frozen at Ϫ80°C until use.
Plasmodium falciparum (3D7) was maintained in human O ϩ erythrocytes as described previously (32). The in vitro anti-plasmodial activity of the rubreserine was determined using the malaria SYBR Green I fluorescence assay as described previously (33,34). In vitro ring-stage intra-erythrocytic P. falciparum parasites (1% hematocrit and 1% parasitemia) were incu-bated with specific concentrations of rubreserine in complete P. falciparum culture medium, with chloroquine disulfate used as a positive control (0.5 M) or vehicle (1ϫ PBS) as a negative control. Fluorescence was measured after 96 h under drug pressure (excitation 485 nm and emission 538 nm). The data, after subtraction of background (chloroquine disulfate-treated infected RBCs and no parasite growth), were expressed as percentage of untreated control to determine cell proliferation.
Rubreserine Preparation-Rubreserine was prepared from (Ϫ)-eseroline fumarate salt (Sigma). Stock solutions of eseroline (10 mM) were made in 50 mM P i (pH 8), 50 mM Tris (pH 8), or 1ϫ PBS (pH 7.5) buffers, depending on the experiment. Under these conditions, eseroline is spontaneously oxidized into rubreserine, a process completed in about 6 h at room temperature. The formation of rubreserine was controlled through the appearance of a characteristic peak of absorption at 475 nm. These stock solutions were stored at 4°C for 48 h or at Ϫ20°C for 1 week. They were serially diluted before use.
Expression and Purification of Recombinant Arabidopsis Enzymes-AtGAT-ADCS and EcADCL proteins were expressed and purified as described previously (19). Arabidopsis cDNAs encoding GMPS, ␤-subunit of AS starting at Ala-51 to remove the plastid transit peptide, and ␣-subunit of AS starting at Ala-61 were amplified by PCR and cloned into the expression vector pET28 (Novagen). E. coli BL21-CodonPlus (DE3)-RIL cells (Stratagene) were transformed and grown using the same protocol as for AtGAT-ADCS. Cells were disrupted, and Histagged recombinant proteins were purified as described previously (19).
Determination of Enzyme Activities-GAT activity can be determined by measuring the production of glutamate and ADCS activity by the production of pABA (19). Standard assays contained 100 mM Tris/HCl (pH 8.0), 5 mM MgCl 2 , 5% v/v glycerol, 0 -0.5 mM L-glutamine, 0 -0.02 mM chorismate, and 9 g ml Ϫ1 of the recombinant plant enzyme. The presence of an excess of EcADCL (20 g ml Ϫ1 , 600 nM) is required for the production of pABA. Reactions (final volume 80 l) were run in 96-well microplates (Greiner), and changes in fluorescence were continuously monitored with a microplate scanning spectrophotometer Safir 2 (Tecan). To monitor glutamate production, an excess of GDH (100 g ml Ϫ1 , 4.2 units ml Ϫ1 ) and 1 mM NAD were added to the assay, and NADH production was followed by its emission at 450 nm (excitation 340 nm). Concentration of pABA was monitored by its fluorescence emission at 340 nm (excitation 290 nm) in the presence of EcADCL.
The activities of GMPS and AS (␤plus ␣-subunits) were measured monitoring glutamate production. In both conditions, the final volume was 500 l. For GMPS, the assay medium contained the following: 200 M XMP, 1 mM ATP, 0.5 mM L-glutamine, 10 g GMPS, 1 mM NAD, 4.2 units ml Ϫ1 GDH. For AS, the assay medium contained the following: 50 M chorismate, 0.5 mM L-glutamine, 10 g each of ␣and ␤-subunits, 1 mM NAD, 4.2 units ml Ϫ1 GDH and various concentrations of rubreserine. The reactions were started by addition of glutamine, and the change in absorbance at 340 nm was monitored with a UV-visible spectrophotometer (Safas).
High Throughput Screening-The screening was conducted at the Center for the Screening of Bio-Active Molecules located at the Commissariat à l'Energie Atomique-Grenoble, France. The library of compounds was purchased from The Prestwick Chemical Library. This library contains 1200 molecules selected for their high chemical and pharmacological diversity. They are marketed and 100% Food and Drug Administrationapproved compounds, supplied in the library at a 10 mM concentration in DMSO. The final concentration used in the enzymatic assay was 100 M, and we verified in separate control experiments that 2% DMSO had no effect on the activity. The enzymatic test used for primary and secondary screenings was identical to the one described above for the GAT activity. Quality of the assay was assessed based on the calculation of the ZЈ factor, as defined by Zhang et al. (35). After optimization of the assay, the ZЈ factor was 0.76 Ϯ 0.08 (a good assay must display a ZЈ factor of Ͼ0.5). The assay was conducted as follows: 50 l of a mixture containing 0.1 M Tris (pH 8), 5 mM MgCl 2 , 5% glycerol, 40 M chorismate, 1 mM NAD, 0.4 unit of GDH, 0.7 g of GAT-ADCS, and 100 M of the various molecules were added in each well (in control wells, molecules were omitted). The reaction was started by injection in each well of 30 l of a solution containing 1.3 mM Gln, and the fluorescence changes (excitation 340 nm and emission 450 nm) were recorded at 0, 15, and 25 min. The most promising molecules were purchased, and their inhibitory properties were manually confirmed.

GAT-ADCS Assay, High Throughput Screening, and GAT-ADCS Inhibition-
The search for enzyme inhibitors by global approaches requires methods to determine the activity that are accurate, robust, and compatible with the automated platforms used for screening of chemical libraries. When coupled with GDH, GAT activity was easily measurable with a UV-visible spectrophotometer (340 nm) or a fluorimeter (excitation 340 nm and emission 450 nm) monitoring NADH accumulation (19,39). Also, it was convenient to use the same procedure to test the inhibitors on GDH alone and to discard molecules that were specific for this last activity. Based on this protocol, we designed a miniature assay easily reproducible and optimized for the identification of active drugs, with a screening coefficient ZЈ (a statistical parameter measuring the quality of the screening assay (35)) of 0.76 Ϯ 0.08 (see "Experimental Procedures"). Because all our attempts to produce active recombinant GAT-ADCS from apicomplexan parasites failed, we used the recombinant plant enzyme as a model for this bifunctional system. We screened the registered Prestwick Chemical Library (1200 compounds) against the GAT activity of the plant GAT-ADCS. After primary and secondary screenings, we identified only four molecules exhibiting IC 50 values that were below 50 M. They were manually tested, and we found that rubreserine (1,3a,8-trimethyl-1,2,3,3a,8,8a-hexahydropyrrolo[2,3-b]indole-5,6-dione, an oxidative product of eseroline (40 -42) that spontaneously forms at alkaline pH ( Fig. 2 and supplemental Fig. S1)) exhibited the best inhibitory properties. Eseroline itself had no effect, indicating that the carbonyl functions resulting from eseroline oxidation were essential for the inhibition. The rate constant of inhibition was rather low (about 0.1 min Ϫ1 ), and a 20-min period of incubation with rubreserine was required to obtain maximal inhibition. We observed GAT-ADCS inhibition independently of the presence or position of the His tag that was added for purification convenience, indicating that such a tag was not involved in the inhibition process. We previously showed that the GAT activity was maximal in the presence of chorismate (the substrate of the ADCS domain) but could also operate independently (19). We observed inhibition of glutamate production in both conditions, i.e. with and without chorismate, and we also observed inhibition of pABA synthesis when GAT-ADCS was coupled with non limiting amounts of ADC lyase (Fig. 3, A and B). In separate control experiments, we verified that rubreserine did not impact ADC lyase activity. In all situations, rubreserine decreased the V m and increased the K m for glutamine, indicating a mixed type inhibition (Fig. 3C) In Vivo Effects of Rubreserine on A. thaliana Growth and Folate Content-When Arabidopsis seedlings were grown on agar plates in the presence of rubreserine concentrations ranging from 50 to 100 M, we observed a dose-dependent growth inhibition (Fig. 4). Interestingly, when the agar plates were supplemented with pABA, the growth was restored for inhibitor concentrations up to 50 M and partially restored for higher concentrations, the maximal effect being obtained with pABA concentrations Ն100 M (Fig. 4). This suggests that for low rubreserine concentrations, at least, the growth inhibition was due to a limitation of pABA synthesis. Because pABA is required for folate biosynthesis, we also measured the potential impact of 5-FTHF on rubreserine-treated plants (Fig. 4). As shown, this folate derivative had similar effect than pABA. Next, we investigated the effect of rubreserine on the folate content. We used Arabidopsis cell cultures for these experiments to dispose of enough material for metabolite determinations. For rubreserine concentrations within the range 25-100 M, the cell division came to an arrest after about 7 days. Similar results were obtained in the presence of sulfanilamide, a well known specific inhibitor of DHPS that blocks pABA utilization in the folate pathway (Fig. 1A). After 7 days, the pools of folate in rubreserine-and sulfanilamide-treated cells (Table 1) decreased by about 40 and 60%, respectively. However, the distributions of folate derivatives were not markedly modified (in all situations, the various representative pools of folates were roughly 50% 5-methyl-THF, 35% 10-formyl-THF plus 5,10methenyl-THF, 8% 5-FTHF, 7% THF plus 5,10-methylene-THF, and 0.3% folic acid (30)). Interestingly, when 100 M pABA were present in the culture medium, the folate content in both control and rubreserine-treated cells were almost identical (Table 1), which was clearly indicative of a protective effect of pABA. Because rubreserine inhibited the pABA branch of the folate pathway, we also attempted to measure the pABA content of Arabidopsis cells. In plants, pABA was found either as free acid or as a glucose ester conjugate. This last form (Ͼ80% of total pABA) is sequestered in vacuoles and does not directly contribute to folate synthesis (44, 45). This makes the determination of free pABA quite difficult because, as previously shown (44), breakdown of the esterified form of folate during sample workup is very difficult to avoid and contributes, sometimes quite significantly, to the amount of free pABA. However, in the presence of 100 M rubreserine, we observed a small but significant decrease of free pABA by about 30% (Table 2). In contrast, the total amount of pABA did not markedly change, suggesting that the glucose ester conjugate in the vacuole and the free acid

TABLE 1 Effect of rubreserine (Rubre) on the folate content of Arabidopsis cells and comparison with sulfanilamide (Sulfa)
Cells were cultivated for 7 days with the various compounds before folate determinations. The total folate contents of control cells cultivated with or without 100 M pABA were 11.8 Ϯ 2 and 10 Ϯ 1.5 nmol g Ϫ1 fresh weight, respectively. Measurements are the average Ϯ S.D. of 4 -6 determinations from two to three independent set of experiments. Results for each set of experiment are expressed as % versus the corresponding control.

Conditions
Total folates  form in the cytosol are not in rapid equilibrium. In sulfanilamide-treated cells, free and total pABA slightly increased by about 30% (Table 2), an expected result taking into account that pABA utilization was blocked (46). Effect of Rubreserine on the Proliferation of Apicomplexan Parasites-Before testing the effect of rubreserine on T. gondii, we first verified with the [3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide] cytotoxicity assay (47) that rubreserine concentrations up to 50 M had no toxicity on confluent HFF cells. Indeed, after 48 h of exposure to 50 M rubreserine, the cell viability was still 95 Ϯ 5% of the control. We then evaluated the effect of rubreserine on Toxoplasma parasites in two different situations as follows: invasion of confluent HFF by the parasites on the one hand, and intracellular development of the parasites within confluent HFF on the other hand. As shown in Fig. 5A, invasion of human cells strongly decreased when the parasites had been previously incubated with rubreserine. This effect was dose-dependent, and the number of intracellular parasites were reduced 2-fold in the presence of 20 M rubreserine (IC 50 ). Likewise, when infected cells were placed in the presence of the inhibitor, the number of parasites in parasitophorous vacuoles decreased after 24 h in comparison with untreated infected cells, which was indicative of a slowing down of the parasite's intracellular division (Fig. 5B). In both situations, rubreserine appeared much more efficient than sulfanilamide and sulfadiazine, and this last sulfonamide drug was widely used in the treatment of toxoplasmosis. To correlate the inhibitory effect with folate biosynthesis, we attempted to determine the folate concentration in T. gondii cells. To our knowledge, there was no report about the intracellular concentration of folates within proliferating T. gondii parasites. We measured folates in parasites grown for 24 h in HFF cells placed in the presence of either 20 or 40 M rubreserine (Table 3). Interestingly, in both conditions the total folate concentration decreased by a factor of 2. Also, similar to what was observed in plants, the distributions of folate derivatives were not significantly changed in the presence of the drug (in these cells the various representative pools of folates were as follows: 18% 5-methyltetrahydrofolate; 25% 10-formyltetrahydrofolate plus 5,10-methenyl-tetrahydrofolate; 42% 5-FTHF; 8% tetrahydrofolate plus 5,10-methylene-tetrahydrofolate; and 7% folic acid). In addition, for a rubreserine concentration of 20 M (IC 50 value), the parasite growth was largely restored in the presence of either pABA or 5-FTHF (Fig. 6A), and the maximal protective effect was obtained for concentrations Ն25 M. Thus, for such inhibitor concentration (20 M), the inhibitory process largely relied on the inhibition of pABA and folate biosynthesis. However, for rubreserine concentrations higher than 20 M, pABA and 5-FTHF had only little effect (Fig. 6B), suggesting that high concentrations of the inhibitor had other non-folatespecific actions. Because antifolate drugs (anti-DHPS and anti-DHFR) are often used in combination, we tested the effect of rubreserine in combination with an anti-DHFR (Fig. 6C). When 20 M rubreserine and 0.4 M pyrimethamine were combined, the inhibition was slightly but significantly increased compared with rubreserine alone. Interestingly, rubreserine alone appeared in our experimental conditions as efficient as a mixture combining 50 M sulfadiazine and 0.4 M pyrimethamine.
Additionally, rubreserine was also tested on the in vitro proliferation of P. falciparum. As shown in Fig. 7, rubreserine exhibited anti-malarial properties because the intraerythrocytic growth of the parasite was strongly inhibited by rubreserine, with an IC 50 of 1 Ϯ 0.04 M (n ϭ 5). Whether folate biosynthesis is also a primary target in these organisms is currently under investigation.
Specificity of Inhibition by Rubreserine-At low rubreserine concentrations, folate biosynthesis appeared as a main target in

Conditions
Total folates Arabidopsis and Toxoplasma. However, at high rubreserine concentrations, the growth activity could not be fully restored by the presence of pABA or 5-FTHF, raising the question of the inhibitor specificity. GAT-ADCS belongs to the family of class I glutamine amidotransferases (18,48), which contains six other members. The GAT domain of GAT-ADCS does not share strong homologies with the GAT domains of the other members of this class, and the best scores were obtained with AS (about 28% identity) GMPS (about 13% identity) and carbamoyl phosphate synthetase II (involved in UMP synthesis, about 15% identity). To test the effect of rubreserine on other members of this group, we attempted to produce these recombinant Arabidopsis activities. We failed to produce active recombinant carbamoyl-phosphate synthetase, but we produced AS (a heterodimeric protein combining the activities of ␤and ␣-subunits, respectively, equivalent to TrpG and TrpE in prokaryotes (49)) and GMPS (a bifunctional enzyme with fused GAT and synthase domains (50), like GAT-ADCS). We determined these activities with the same GDH-coupled assay that we used for GAT-ADCS. As shown in Table 4, rubreserine also inhibited GMPS and AS activities, although inhibition of GMPS required higher concentrations of inhibitor. Toxoplasma, but not Plasmodium, is an auxotroph for tryptophan because of the lack of AS activity (51,52). However, the enzymes involved in the synthesis of GMP (GMPS, a bifunctional enzyme, as in plants (53)) and UMP (carbamoyl-phosphate synthetase II, a heterodimer in plants (54), but a single protein in apicomplexan   parasites (55)) was present in these organisms. When a mixture containing pABA, anthranilate, UMP, and GMP (50 M each) was added to the culture medium of rubreserine-treated plants or was present in the proliferation assay of Toxoplasma, the growth recovery for both organisms was not markedly improved compared with that obtained with pABA or 5-FTHF alones (see Fig. 6B for the experiments with T. gondii).

DISCUSSION
Two main conclusions can be drawn from this study: first, our data validate for the first time the use of the bifunctional GAT-ADCS as an efficient drug target in eukaryotic cells, and second, we identified a new scaffold that inhibits plant growth and proliferation of apicomplexan parasites.
The screening test we used appeared efficient to select active compounds from Prestwick chemical library, and we found that an oxidative product of eseroline, rubreserine, was inhibiting the GAT activity of AtGAT-ADCS. Eseroline and rubreserine were identified a long time ago as metabolites of physostigmine (eserine), an alkaloid present in the Calabar bean (Physostigma venenosum), and previously used for its potent anticholinesterase activity. To our knowledge, rubreserine and eseroline have no current use in any medical purpose. Pharmacological studies indicated that these two molecules were poor inhibitors of cholinesterase (56), but there was no reports indicating that rubreserine could affect folate biosynthesis and inhibit the growth of plants and the proliferation of parasites. Thus, we describe for this compound a new biological effect with interesting therapeutic potentialities.
When GAT-ADCS was coupled with ADC lyase, the apparent constant of inhibition (K i ) for pABA formation was estimated to be Ͻ10 M, which is the best constant of inhibition obtained so far for the biosynthesis of pABA. How rubreserine affects the protein activity is not yet understood, however, and is currently under investigation. The obvious difference between the chemical structures of eseroline and rubreserine is the presence of two carbonyl functions on the aromatic ring of the latter compound. Thus, these carbonyl functions were presumably at the origin of the inhibitory effect. It must be noted that many other natural quinonoid compounds were shown to display antimalarial properties (57), although the targets and modes of action were not described for most of them.
The rubreserine-dependent growth inhibitions of Arabidopsis and Toxoplasma were specifically associated with an inhibition of folate biosynthesis for concentrations ՅIC 50 , although higher concentrations might also inhibit other activities. The correlation between rubreserine and folate biosynthesis was observed by direct and indirect approaches. The direct approach indicated that Arabidopsis cells exhibited folate and free pABA contents lowered by 45 and 25%, respectively, in the presence of the inhibitor. In addition, when pABA was present in the culture medium of rubreserine-treated plant cells, the folate level was almost identical to the control, indicating a protective role of pABA. Likewise, the folate level in Toxoplasma cells was two times lowered by rubreserine. The impact of rubreserine on folate biosynthesis was also shown indirectly. Indeed, for rubreserine concentrations close to the IC 50 values, the rubreserine-dependent growth inhibitions of Arabidopsis and Toxoplasma were for a large part reversed by the addition of pABA or 5-FTHF, indicating that the biosynthesis of pABA displayed a particular sensitivity to rubreserine and that the resulting decrease of folate biosynthesis contributed to the inhibitory process. The mode of action of rubreserine in Plasmodium is currently under investigation to determine to what extent folate biosynthesis is inhibited in this organism and the contribution of such an inhibition to the whole inhibitory process.
It is interesting to compare the effects of GAT-ADCS and DHPS inhibitors because both types of drugs impact pABA metabolism. Interestingly, rubreserine appeared in our experimental conditions much more efficient against T. gondii than sulfadiazine, a sulfonamide widely used for the treatment of severe toxoplasmosis. Indeed, the IC 50 value calculated for rubreserine was significantly lower than the IC 50 for sulfadiazine estimated from this study (Ͼ50 M). The IC 50 for sulfadiazine may fluctuate widely depending on the parasite strain, from about 20 M to more than 200 M, and is generally Ͼ30 M (58). Such a variation presumably illustrates the occurrence of resistance within the numerous Toxoplasma strains. Sulfonamide drugs are normally not used alone against parasites of the Apicomplexa phylum because of their limited activity (8). However, they are potent synergizers of DHFR inhibitors (exemplified in Fig. 6B), which is the reason why these molecules are used in combination. Indeed, inhibition of DHPS decreases the de novo synthesis of dihydropteroate, which, in turn, leads to reduction of dihydrofolate, the substrate of DHFR. Because the amount of dihydrofolate is decreased, the efficiency of DHFR inhibitors increases, and lower doses of these toxic molecules are required. When rubreserine was used in combination with pyrimethamine in Toxoplasma, we also observed a small but significant synergistic effect, and such an association appeared as efficient as a mixture combining sulfadiazine and pyrimethamine. Thus, molecules with a hexahydropyrrolo[2,3-b]indole-5,6-dione scaffold, such as rubreserine, could be interesting structures to develop novel drugs that could represent alternatives to sulfonamides.