The Glutathione Biosynthetic Pathway of Plasmodium Is Essential for Mosquito Transmission

Infection of red blood cells (RBC) subjects the malaria parasite to oxidative stress. Therefore, efficient antioxidant and redox systems are required to prevent damage by reactive oxygen species. Plasmodium spp. have thioredoxin and glutathione (GSH) systems that are thought to play a major role as antioxidants during blood stage infection. In this report, we analyzed a critical component of the GSH biosynthesis pathway using reverse genetics. Plasmodium berghei parasites lacking expression of gamma-glutamylcysteine synthetase (c-GCS), the rate limiting enzyme in de novo synthesis of GSH, were generated through targeted gene disruption thus demonstrating, quite unexpectedly, that c-GCS is not essential for blood stage development. Despite a significant reduction in GSH levels, blood stage forms of pbggcs 2 parasites showed only a defect in growth as compared to wild type. In contrast, a dramatic effect on development of the parasites in the mosquito was observed. Infection of mosquitoes with pbggcs 2 parasites resulted in reduced numbers of stunted oocysts that did not produce sporozoites. These results have important implications for the design of drugs aiming at interfering with the GSH redox-system in blood stages and demonstrate that de novo synthesis of GSH is pivotal for development of Plasmodium in the mosquito. Copyright: ß 2009 Vega-Rodríguez et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.


Introduction
Plasmodium infection leads to increased oxidative stress in both the vertebrate and mosquito hosts. The high proliferation rate of parasites results in the production of large quantities of toxic redox-active by-products. Reactive oxygen species (ROS) are generated within the infected RBC (iRBC) as a result of degradation of hemoglobin in the food vacuole of the parasite [1,2]. In addition, ROS arise from the production of nitric oxide and oxygen radicals produced by the host's immune system in response to iRBC bursting and merozoite release [1,3]. In the Anopheles mosquito vector, nitric oxide species and ROS are produced in response to invasion of midgut epithelial cells by the parasite [4][5][6] suggesting the requirement of efficient defense mechanisms to protect against oxidative damage. A detailed study of the Plasmodium genome reveals the absence of genes encoding the antioxidant enzymes catalase and glutathione peroxidase [7,8]. The lack of a glutathione peroxidase gene raises doubts about the relevance of the glutathione (GSH) pathway in detoxification of oxidative stress in Plasmodium. However, supportive of a role for GSH metabolism in the detoxification process are the observations that the P. falciparum glutathione S-tranferase enzyme, which conjugates GSH to other molecules via the sulfhydryl group, displays peroxidase activity [9]. The Plasmodium GSH pathway, in conjunction with the thioredoxin redox system, could indeed act as a primary line of defense against oxidative damage [10].
To date, the role of the Plasmodium GSH antioxidant system has only been studied in the context of the erythrocytic stages [2,8]. GSH is a thiol-based tripeptide implicated in a variety of cellular processes, including detoxification of xenobiotics and protection against ROS [11,12]. Additional roles ascribed to GSH based on biochemical studies in Plasmodium iRBC include serving as cofactor for enzymes such as glutathione-S-transferase and as reducing agent for ferriprotoporphyrin IX, the toxic by-product of hemoglobin digestion [13]. Evidence has been presented that P. falciparum does not utilize GSH from the host RBC since the parasite membrane is neither permeable to host GSH nor cglutamylcysteine [14,15]. Plasmodium is therefore thought to be dependent on its own GSH biosynthetic pathway. GSH is synthesized in Plasmodium by consecutive reactions facilitated by the enzymes c-glutamylcysteine synthetase (c-GCS) and glutathione synthetase (GS), independently of GSH biosynthesis in the host RBC, which becomes inactive after invasion [12,14,[16][17][18]. However, Platel et al. [19] hypothesized that host GSH can be transported into the P. berghei food vacuole via hemoglobincontaining endocytic vesicles, based on data showing that GSH can detoxify the toxic ferriprotoporphyrin IX inside the parasite's food vacuole.
c-GCS catalyzes the rate limiting step during GSH biosynthesis [20] and is inhibited in both P. falciparum and P. berghei by the generic c-GCS inhibitor L-buthionine-(S,R)-sulphoximine (BSO) resulting in reduced GSH levels and, depending upon BSO concentration, in parasite death [19,21]. These results are consistent with the expectation that de novo synthesis of GSH by Plasmodium is essential for parasite development within the iRBC. Given the oxidative environment of the iRBC it has been proposed that enzymes involved in parasite GSH biosynthesis are promising targets for the development of novel antimalarial agents [1,2,21].
In this study we analyzed the GSH biosynthetic pathway using reverse genetics. Following targeted gene disruption of the single copy gene encoding c-GCS in P. berghei (pbggcs) we assessed the role and essential nature of de novo synthesis of GSH in parasite growth and development within the RBC and the mosquito midgut. Unexpectedly, c-GCS is not required for blood stage development but de novo synthesis of GSH appears to be essential to complete sporogonic development of the parasite in the mosquito vector. These observations have important implications for the design of drugs aiming at interfering with the GSH redox-system in blood stages and may help to further unravel the functional role of GSH during development of malaria parasites in the mosquito.

Results
Disruption of the c-GCS-encoding pbggcs gene does not prevent blood stage development To examine the role of c-GCS in P. berghei growth and development, the gene encoding c-GCS (pbggcs PB001283.02.0) was disrupted using standard genetic modification technologies. In two independent experiments (exp. 866 and 985), parasites of the reference line 507cl1 (P. berghei ANKA strain expressing GFP; ANKA wt GFP) were transfected with the DNA-construct aimed at disruption through double cross-over integration (Fig. 1A). One parasite clone from each transfection experiment (pbggcs 2 1; pbggcs 2 2) was analyzed. Correct integration of the constructs into the parasite genome was confirmed by Southern analysis of digested DNA (Fig. 1B) and field inverted gel electrophoresis (FIGE) separated chromosomes (results not shown). The absence of pbggcs mRNA in blood stages of pbggcs 2 1 and pbggcs 2 2 was demonstrated by PCR analysis of cDNA and Northern analysis of mRNA of blood stage parasites ( Fig. 1C and 1D).

P. berghei blood stages can grow and develop in vivo in spite of low levels of intracellular GSH
Given that c-GCS catalyzes the rate-limiting step in GSH de novo biosynthesis, we predicted that disruption of the pbggcs locus would abolish parasite GSH levels. Parasite extracts were analyzed by HPLC for total GSH concentrations. Glutathione levels were significantly reduced (P,0.001) in blood stages of both pbggcs 2 1 (1.0 nmol/10 9 parasites, n = 16) and pbggcs 2 2 (0.2 nmol/10 9 parasites, n = 22) as compared to wild type parasites (7.4 nmol/ 10 9 parasites, n = 20) (Fig. 2). Thus, the disruption of pbggcs, caused a significant reduction but incomplete depletion of GSH. These results demonstrate that P. berghei blood stages can grow and develop in vivo with minimal levels of GSH. The much higher concentrations found in wild type parasites are not required for survival in the RBC.
Disruption of pbggcs has a minor effect on blood stage proliferation When the mutant parasite lines were cloned by limiting dilution, we observed a minor decrease in the growth rate of the mutant parasites as compared to wild type. After injecting mice with a single wild type parasite, parasitemias reached 0.5-2% at 8 days post infection (d.p.i.) (mean = 8.0; st.dev. = 0; n = 6). The same level of parasitemia was reached by parasites of the mutant lines one day later (mean = 8.7; st.dev. = 0.5; n = 6). These results suggest a delay in growth of pbggcs 2 parasites as compared to wild type. To further examine the potential effect of the absence of c-GCS on growth rate, the course of parasitemia in groups of mice infected with 2610 2 or 2610 3 mutant or wild type parasites was followed in a second set of experiments (Fig. 3). The rate and extent of parasite multiplication was reduced in the mutant parasites. Therefore, although c-GCS is not essential for growth and multiplication of blood stage Plasmodium in vivo, the absence of c-GCS expression affects parasite growth. c-GCS is essential for mosquito stage development of malaria parasites To date, Plasmodium GSH synthesis and metabolism have been studied only during blood stage development. To evaluate the effect of pbggcs disruption on the development of mosquito stages, Anopheles stephensi mosquitoes were infected with pbggcs 2 parasites and total ookinete numbers present in the mosquito midgut were determined 24 hrs post infection (h.p.i.). Ookinete numbers were reduced by 30-40% as compared to wild type parasites (n = 25) in both mutant clones (pbggcs 2 1: 31%, P = 0.056, n = 25; pbggcs 2 2: 38%, P = 0.043, n = 25) (Fig. 4A).
Oocyst development was examined at 2 days p.i. to determine if pbggcs 2 ookinetes were able to cross the mosquito midgut wall and develop beneath the basal lamina. We used antibodies against Pbs21, a protein expressed on the surface of ookinetes and young oocysts to assess early oocyst number and morphology (Fig. 4B). Pbs21-positive oocysts counts revealed that the median oocyst numbers were reduced by 70% in pbggcs 2 1 (P,0.0001, n = 40) and by 95% in pbggcs 2 2 (P,0.0001, n = 40) when compared to wild type parasites. Oocysts of both wild type and pbggcs 2 parasites showed a similar distribution of Pbs21 (Fig. 5A). Analysis of the number of mature mercurochrome stained oocysts at day 12 p.i.

Author Summary
The antioxidant systems of malaria parasites (Plasmodium spp.) are potential targets for the development of antimalarials. The glutathione (GSH) redox system constitutes one of the Plasmodium primary lines of defense against damage caused by reactive oxygen species and other forms of chemical stress. GSH is synthesized de novo by the sequential action of gamma-glutamylcysteine synthase (c-GCS) and GSH synthase (GS). Biochemical studies have suggested that parasite survival depends on functional de novo GSH synthesis. Using reverse genetics we interrupted the GSH biosynthetic pathway in the rodent malaria Plasmodium berghei by disrupting the pbggcs gene. The mutation caused minor changes in parasite growth in the mammalian host but development in the mosquito was completely arrested at the oocyst stage. These results suggest that the GSH biosynthetic pathway, while essential for mosquito stage development, is not an appropriate target for antimalarials against blood stages of the parasite.
showed that median oocyst numbers were reduced by 65% in pbggcs 2 1 (P,0.0001, n = 47) and by 84% in pbggcs 2 2 (P,0.0001, n = 68) (Fig. 4B). This comparable reduction in oocyst numbers at days 2 and 12 p.i. indicates that the disruption of the pbggcs gene results in an early effect on oocyst development, most probably by reducing the number of ookinetes that reach the basal lamina and successfully transform into oocysts. Analysis of the mercurochrome stained pbggcs 2 oocysts at day 12 p.i. revealed the absence of internal structures typical of wild type parasites during sporozoite formation. In addition, these oocysts showed a significant size reduction (Fig. 5C).
The capsule protein CAP380 [22] and the circumsporozoite protein CS [23], expressed during mid to late oocyst development and sporozoite formation, respectively, were assessed in the pbggcs 2 oocysts by immunofluorescence analysis (IFA). Antibodies to both proteins stained pbggcs 2 oocysts at day 12 p.i. Reactivity with anti-PbCap380 revealed normal formation of the oocyst capsule and confirmed the reduction in the size of pbggcs 2 oocysts previously observed after mercurochrome staining ( Fig. 5B and 5C). The reactivity pattern of pbggcs 2 oocysts with PbCS was clearly different from that of wild type oocysts (Fig. 5B). The normal stippling staining characteristic of the formation of sporozoites was absent in pbggcs 2 oocysts. The lack of internal structures in mercurochrome-stained oocysts, the absence of the characteristic sporozoite staining pattern with anti-CS antibodies, and the reduction in size indicate a defect in oocyst development and sporozoite formation.
It is known that mitochondria are especially sensitive to oxidative stress and that reduced GSH levels can affect the function of mitochondria as a result of oxidative damage [24,25]. To probe mitochondria integrity, oocysts were stained at day 12 p.i. with MitoTracker, a dye which detects an intact mitochondrial membrane potential. A distinct and bright staining pattern was observed in wild type oocysts (Fig. 6A). The stippled staining pattern represents the multiple mitochondria of individual sporozoites developing within the oocyst. In contrast, only a small proportion of the pbggcs 2 oocysts were stained with MitoTracker ( Fig. 6C and 6D) and the few MitoTracker stained oocysts, showed an aberrant staining pattern characterized by a diffused staining with reduced intensity compared to the wild type oocysts.
To further analyze the morphology of the pbggcs 2 oocysts, thinsection transmission electron microscopy was performed on oocysts 12 days p.i. (Fig. 7). Compared to wild type, most of the pbggcs 2 oocysts exhibited a pattern of necrosis with degenerated organelles dispersed in an electron-lucent cytoplasm. Sparse mitochondria were visible in contrast to wild-type, characterized by an increased amount of mitochondria at the onset of sporozoite formation. Rough endoplasmic reticulum was abundantly present in wild type parasites, whereas the density of this organelle in pbggcs 2 oocysts was strongly reduced. Also the nuclei of pbggcs 2 oocysts were sparse and pycnotic. Despite the highly aberrant morphology, the pbggcs 2 oocysts showed a capsule resembling that of wild type oocysts, which is in agreement with IFA data (Fig. 5B). However, in some sections of pbggcs 2 oocysts, heterogeneities in the thickness of the capsule were observed (Fig. 7) whereas in wild type oocysts this structure was more homogenous. Together, these data reveal that the development of parasites lacking the expression of c-GCS is halted during development of the oocysts at the onset of sporozoite formation.
To confirm that pbggcs 2 oocysts are incapable of producing viable sporozoites, salivary glands from infected mosquitoes were dissected and analyzed on days 18-25 p.i. (data from 25 days p.i. not shown). None of the salivary glands from mosquitoes infected with pbggcs 2 parasites (n = 20) contained sporozoites (Fig. 4C). Moreover, feeding infected mosquitoes on naïve mice at day 20 p.i. did not result in infection (data not shown). Thus, oocyst development in parasites lacking expression of c-GCS is halted, preventing the development of viable sporozoites and interrupting transmission.  Representative calibration curve using GSH standards in a range between 4 and 24 mM. (C) Comparison of total GSH concentration of WT, pbggcs 2 1, pbggcs 2 2, and pbggcs-comp parasites. GSH levels were determined in extracts of purified blood stages obtained from mice with asynchronous infections. A One Way ANOVA test shows statistically significant differences in GSH concentration between pbggcs 2 parasites and wild type (P,0.0001) and between pbggcs 2 and pbggcs-comp parasites (P,0.0001). Numbers on boxes represent the mean concentration of GSH. Bars represent the standard error of the mean. Wild type n = 20, pbggcs 2 1 n = 16, pbggcs 2 2 n = 22, pbggcs-comp n = 16. doi:10.1371/journal.ppat.1000302.g002 Genetic complementation of pbggcs 2 parasites restores parasite development in the mosquito We used a reverse genetics approach to complement pbggcs 2 parasites with a wild type copy of pbggcs. The pbggcs 2 parasites (pbggcs 2 2) were transfected using a construct that contains the pbggcs gene under the control of the strong, constitutive eef1a promoter (Fig. 8A) [26]. In addition, this plasmid contains the human dihydrofolate reductase (hdhfr) selection cassette and the D-type small subunit rRNA (dssurrna) as targeting sequence for integration into the parasite genome by single cross-over recombination. Resistant parasites were selected by treatment of mice with WR99210 [27]. Correct integration of the complementation plasmid into the dssurrna genomic locus of these parasites (pbggcscomp parasites) was confirmed by Southern analysis of contour clamped homogeneous electric field (CHEF) separated chromosomes (Fig. 8B). Detection of pbggcs mRNA by PCR analysis of cDNA from pbggcs-comp parasites confirmed successful expression of the introduced pbggcs gene in blood stage forms (Fig. 8C).
A. stephensi mosquitoes were allowed to feed on mice infected with pbggcs-comp parasites. Mosquito midguts were dissected 12 days p.i. and oocysts morphology was analyzed by light microscopy following mercurochrome staining. As shown in Fig. 8D, in pbggcs-comp parasites the morphology of oocysts was similar to that of wild type, both with respect to size and the formation of sporozoites within oocysts. The complemented parasites displayed increased GSH levels (9.1 nmol/10 9 parasites, n = 16) when compared to pbggcs 2 parasites, demonstrating successful restoration of GSH biosynthesis to levels equivalent to those of wild type parasites (Fig. 2). When mosquitoes infected with pbggcs-comp parasites were allowed to feed on naïve mice, mice became infected (data not shown). Analysis of the genotype of pbggcs-comp parasites after mosquito transmission showed only the complemented genotype, confirming that passage to the mammalian host is only possible in the presence of pbggcs expression (Fig. 8B). In summary, these results show that the complementation with a functional copy of pbggcs restored the development of oocysts and sporozoites of the pbggcs 2 parasites and GSH synthesis and confirm that the arrested oocyst development phenotype and the blocking of transmission observed for the pbggcs 2 parasites is the result of the disruption of the pbggcs gene.

Discussion
The data presented herein conclusively demonstrate that P. berghei parasites do not require the de novo synthesis of GSH for asexual development in mice. These results contrast the observed growth inhibition of P. falciparum in vitro following administration of the generic c-GCS inhibitor BSO [21] and in vivo in P. berghei [19]. The highly significant reduction in parasite GSH levels resulting from disruption of the pbggcs gene caused only a relatively minor effect on the growth rate of intraerythocytic stages. Perhaps the low GSH levels still detected in the pbggcs 2 parasites are sufficient to support growth within the RBC. This GSH could be derived from the iRBC and transported to the parasite food vacuole via hemoglobin-containing endocytic vesicles, as proposed by Platel and others [19]. Evidently, GSH levels found in blood stages of wild type parasites are not essential for survival.
It could be argued that c-GCS is not the only target of BSO in P. berghei, as has been proposed for Trypanosoma brucei [28], where supplementary GSH did not prevent death as a result of BSO treatment but was able to rescue the lethal effects of a c-GCS gene knockdown. Although the molecular basis for strain dependent sensitivity of blood stage P. falciparum to BSO is unknown [29], our findings might lead one to anticipate a difference in dependence on the de novo GSH synthesis of blood stages of different species of Plasmodium. Plasmodium vivax and P. berghei, for example, preferentially infect reticulocytes [30], which contain higher levels of GSH . Antibody-stained oocysts were detected by indirect fluorescence microscopy using Rhodamine Red TM -X goat anti-mouse IgG (H+L) or Texas RedH-X goat anti-rabbit IgG (H+L) secondary antibodies (Invitrogen), respectively. The pbggcs 2 oocysts produced the PbCap380 protein but are strongly reduced in size compared to wild type (see also C). Anti-CS staining shows that, although CS is produced, 12 day old oocysts lack the typical CS staining pattern of wild type oocysts. (C) Oocyst size on dissected midguts 12 d.p.i. doi:10.1371/journal.ppat.1000302.g005 than normocytes [31]. The c-GCS deletion mutants should provide further insight into the role of the de novo synthesis of GSH and its ability to serve as a viable drug target to prevent blood stage development.
Although our results show that blood stages can survive and multiply without expression of c-GCS, we observed a clear growth delay of the blood stages of pbggcs 2 parasites. This growth delay may be the result of a prolonged cell cycle, the production of fewer daughter merozoites within the schizonts, reduced invasion efficiency, or reduced proliferation in normocytes as compared to reticulocytes. Interestingly, a longer cell cycle has been reported for Saccharomyces cerevisiae and T98G glioblastoma cells that were depleted of GSH [32,33]. This growth delay was detected during the G1 and S phase transition, when cells prepare for DNA replication, and it has been suggested that the slower cell cycle could provide sufficient time to repair the damage caused by the increased oxidative stress after GSH depletion. Likewise, a prolonged cell cycle as a response to decreased GSH-levels could endow Plasmodium blood stages with enough time to repair any critical damage as a consequence of a reduced GSH concentration. However, because GSH in the cell has additional functions apart from its role as a general thiol redox buffer, further detailed studies on cell cycle, invasion of and survival in erythrocytes of different ages are needed for more insight into the mechanism underlying the reduced growth rate of P. berghei blood stages as a result of low GSH levels.
In contrast to the relatively minor effect in blood stage development, disruption of pbggcs had a dramatic effect on parasite growth in the mosquito, completely blocking transmission. The reason for this differential requirement for endogenously-sourced GSH may lie in the potential for the profound drop in parasite GSH levels to be replaced by reducing agents from either the parasite or the iRBC. However, comparable compensatory sources might not be available in the mosquito. Furthermore, in the midgut, Plasmodium is initially extracellular. Levels of reactive oxygen species (ROS) are known to increase in both the midgut and hemolymph of Anopheles gambiae mosquitoes following a P. berghei-infected blood meal, as part of the innate immune response [6,34]. Parasites lacking one of the arms of its redox defense might be less able to survive insect immunity. Alternatively, in the absence of GSH de novo synthesis, mosquito stages of Plasmodium may simply fail to deal with the intrinsic increase in oxidative stress associated with the endogenous accumulation of ROS by-products of cellular metabolism during ookinete and sporogonic development. In Plasmodium gametocytes and ookinetes, the number of mitochondria increases, their structure is changed, and their activity is intensified [35][36][37]. Mitochondria are one of the main ROS producers in the cell, therefore one of the first organelles to be adversely affected by a poorly restrained increase in these harmful radicals. The damage that ROS can induce in this organelle, which include mitochondrial DNA mutations as well as lipid and protein peroxidation, is limited by GSH [25]. The significant reduction of GSH in pbggcs 2 mutants could result in accelerated mitochondrial damage and loss of function, further production of ROS and ultimately, death by apoptosis [38]. Therefore, the augmented activity of mitochondria during the development of P. berghei ookinetes and oocyst might lead to an increased dependence on GSH for maintaining ROS reduction and redox homeostasis.
Interestingly, it has been shown using reverse genetics that disrupting genes encoding other proteins associated with the parasite redox systems, such as thioredoxin peroxidase from P. falciparum (pftpx-1) and P. berghei (pbtpx-1), are also not essential for blood stage development [39,40]. In addition, depletion of the antioxidant plasmoredoxin, an enzyme from the thioredoxin superfamily, which links the thioredoxin and the glutathione redox system [41], showed no detectable effect on parasite development in the RBC [42]. The redundancy in function of multiple members of the thioredoxin family has been postulated as a possible explanation for these results [42]. Furthermore, functional redundancy stemming from the presence of both thioredoxin and glutathione redox systems as a defense against oxidative stress might also explain the fact that the individual systems are not essential for asexual development [41][42][43]. This may be comparable to the multiple redundant pathways for hemoglobin digestion [44] and erythrocyte invasion [45] developed by the parasite to optimize growth and survival within the host RBC.
The observation that fully developed infectious pbggcs 2 sporozoites were not produced and that infected mosquitoes were unable to transmit the infection to naïve mice are extremely relevant to the field. Disruption of the pbggcs gene resulted in attenuated development of oocysts and subsequent failure to produce sporozoites, as evidenced by the production of an oocyst capsule that is highly heterogeneous as to thickness, the loss of mitochondria membrane potential in the mutant oocysts, and abnormal patterns of expression of the circumsporozoite (CS) protein [23]. A significant reduction in the amount of mature pbggcs 2 ookinetes in the mosquito midguts compared to wild type was also detected followed by an even greater reduction in numbers of oocysts at the basal lamina. The combined results from mosquito stages of pbggcs 2 parasites point to a continuing failure of the pbggcs 2 mutant to develop in the vector. This spectrum of defects is quite consistent with a mutant that has a reduced capacity to deal with ROS due to an absence of de novo synthesis of GSH. In agreement with the present findings, both protracted and altered cell cycles have been reported for S. cerevisiae, Arabidopsis thaliana and T98G glioblastoma cells when available GSH levels were critically depleted either through drug treatment or genetic manipulation [32,33,46]. Further studies are needed to determine whether a comparable cell cycle defect underlies the deleterious effects of the absence of c-GCS in Plasmodium mosquito stages. The fact that blood stage parasites could survive without c-GCS raises some doubts on its potential use as a target in the development of new drugs against parasite blood stages in man. A clarification of such issues requires resolution of the precise roles of GSH, its biosynthetic pathway and the interplay with other redox systems. The role of the parasite GSH-redox system in mosquito stages has not been previously addressed and our results encourage further studies aimed at unraveling the critical role of this pathway for the survival of the parasite in its vector.   cloned into the TOPO TA vector (Invitrogene) and then sequenced to obtain pbggcs-TOPO plasmid. To generate the 39 targeting region, a fragment of 754 bp was digested from pbggcs-TOPO plasmid with the HincII/ NotI enzymes and cloned into EcoRV/NotI pL0001-59pbggcs digested plasmid to create the pbggcs disruption vector, pL1217. For transfection, pL1217 was linearized with KpnI/SacII restriction enzymes and transfected into P. berghei purified schizonts of line 507cl1 [47]. Briefly, this line, ANKA wt-GFP, expresses GFP under the control of the constitutive eukaryotic elongation factor 1A (eef1aa) promoter and has been selected by Fluorescence Activated Cell Sorter (FACS) sorting based on GFP expression as described by Janse et al. [47]. Transfection, selection and cloning of pbggcs 2 parasites was carried out as previously described [27]. Correct integration of pL1217 into the pbggcs genomic locus was confirmed by standard Southern blot analysis of digested genomic DNA using tgdhfr/ts and pbggcs specific probes (Fig. 1B). The 921 bp tgdhfr/ts probe was obtained by digesting pL0001 with SalI and the 833 bp pbggcs-specific probe was obtained by digesting the pbggcs-TOPO plasmid with SpeI/ NheI. Hybridizations were performed according standard methods. Expression of the pbggcs gene was analyzed by PCR amplification of reverse transcribed mRNA. Total RNA of blood stages obtained from mice with asynchronous infections was isolated using RNA STAT-60 (Tel-Test Inc.) according to manufacturer's specifications. Complementary DNA (cDNA) was synthesized with the Superscript II RNase H-Reverse Transcriptase (Invitrogen) following manufacturer's recommendations. PCR's were carried out on 1 ml of synthesized cDNAs using the pbggcs specific primers F1822 (59-TTAACGGTTTTCTG-TAAATGC-39)/R2536 (59-TTCTTCTTATTTTCATACAAT-GCTC-39) which amplifies 746 bp of the 59 region of the pbggcs gene including the 59UTR. A control PCR was included to exclude potential contamination with gDNA by using primers directed to the two exons of the P. berghei hepatocyte erythrocyte protein 17 gene (pbhep17) homologue of the pyhep17 [48].

DNA constructs and parasite transfection
Plasmid pL0009 (MR4) was used as a backbone for the pbggcscomplementation construct. This plasmid contains the hdhfr selectable marker which confers resistance to pyrimethamine and the antimalarial WR99210 and targeting sequence for integration into the cor d-ssurrna by single cross-over recombination. First, the green fluorescence protein (gfp) mutant 3 gene (BamHI/XbaI) of pL0017 was exchanged for the e-gfp gene of plasmid pEGFP-NI (Clonetch, subcloned SacII/NotI in pBluescript-SK) to create the ef-eGFP vector. Then, the coding sequence of the pbggcs gene including the 39 UTR of pbggcs-Topo plasmid was digested with KpnI/NcoI and cloned into the ef-eGFP plasmid to obtain pL1136. Finally, the EcoRV fragment of pL1136 plasmid (which contains the pbggcs expression cassette) was cloned into the pL0009 vector to obtain the pbggcs complementation plasmid (pbggcs-comp). Both expression cassettes (pbggcs and hdhfr) are in the same orientation in this vector. Plasmid pbggcs-comp was linearized with SacII and transfected into pbggcs2 2 parasites as previously described [27]. Transfected parasites were selected after a four day treatment with 16 mg/kg body weight of the drug WR99210. Correct integration of the complementation construct into the c/d-ssurrna on chromosome 5/6 was verified by Southern analysis of chromosome separated by Contour-clamped Homogeneous Electric Field (CHEF) electrophoresis using standard procedures. Southern analysis was performed using the pbggcs and tgdhfr/ts specific probes described above. Expression of the pbggcs gene in pbggcs-comp parasites was analyzed by reverse transcription PCR (RT-PCR) and Northern blot analysis of total mRNA of blood stage parasites. RT-PCR analysis was done using the pbggcs specific primers PIY-F (59-TATAAAGATGTAAATACAG-39)/DAM-R (59-CATTCCAAAAAACATTGCATC-39). Northern blots were hy-bridized to the pbggcs-specific probe used in the Southern analysis, as described above. As a loading control the Northern blot was hybridized with a ribosomal RNA probe (primer L647) [49].

Mice
Random-bred Swiss albino CD-1 female mice (Charles River Laboratories, Wilmington, MA, USA), 6-8 weeks old, weighting 20 to 35 g at the time of primary infection were used throughout the study. They were kept in a room with a temperature of 22uC and a 12h light /12h dark cycle. All studies involving laboratory animals were performed in accordance with the regulations of the US Institutional Animal Care and Use Committee (IACUC) and the regulations of the Dutch Experiments on Animals Act.

Determination of GSH levels in blood stage parasites
Parasite intracellular GSH levels were determined using a modified HPLC method previously described [modified from 50 and 51].
Parasite isolation. P. berghei infected blood (0.8-1.0 mL) was harvested by heart puncture from mice with parasitemias between 5% and 15%. White blood cells and platelets were removed by using a WhatmanH CF11 cellulose column [52]. Infected RBCs were centrifuged and washed with saline solution and subsequently lysed with saponin (0.15%) for 10 minutes at 4uC. Free parasites were centrifuged, washed with saline solution and resuspended at a final concentration of 50610 6 /100 mL in a buffer containing 3.5 mM MgCl 2 , 110 mM KCl, 40 mM NaCl, 20 mM Hepes, 6 mM EDTA pH 7.4 and a protease inhibitors cocktail [53]. Prior to total GSH extraction, parasite lysis was achieved by three successive freeze (liquid nitrogen) and thaw (37 uC) cycles.
GSH extraction. Parasite extracts were treated with 12.5 mM dithioerythritol (DTE) for 5 minutes to reduce total GSH, which include GSH, GSSG and GS-protein in the sample. Phosphoric acid (1.0 M) was then added for protein precipitation. After centrifugation, 150 mL of the sample were transferred to a 1.5 mL tube containing 100 mg of sodium bicarbonate. To this solution, one hundred microliters (100 mL) of 0.1 M Tris-HCl pH 8.00 and 2 mL monobromobimane (MBrB) were added. The derivatization was carried out at room temperature for 5 minutes in the dark and was terminated by adding 20 mL of 60% phosphoric acid. Samples were centrifuged at 14, 000 RPMs for two minutes and the supernatant was filtered (Millipore 0.22 mm filter) at 5,000 RPMs for two minutes.
Total GSH measurement. Samples were loaded on to a Hewlett Packard HP ODS Hypersyl column (5.00 mm, 20064.0 mm) and analyzed as MBrB derivatives with a Hewlett Packard 1050 Series HPLC. The compounds were eluted at a flow rate of 1.2 mL/min with mobile phase (0.25% acetic acid, 0.25% phosphoric acid, 9% acetonitrile, and 90.5% water). After 7 minutes, a gradient was established to 5% mobile phase and 95% acetonitrile up to 8.5 min and the flow rate was increased to 1.7 mL/min for 0.5 minutes. At 9 minutes, the system was returned back to 100% mobile phase and 1.2 mL/min. Total GSH in the effluent was measured using a Hewlett Packard 1046A Programmable fluorescence detector (excitation 388 nm; emission 491 nm). Fluorescence intensity was measured and the area of the GSH peak (not peak height) was used to quantify the amount of total GSH when compared to the calibration curve. GSH calibration curves were ran every time a set of samples were analyzed using six different concentrations between 4 and 24 mM GSH.

Determination of growth of asexual blood stages in vivo
The asexual multiplication rate in vivo was analyzed by determination of the parasitemia at day 7-9 after injecting mice (i.v.) with a single parasite (during the procedure of cloning of the mutants by limiting dilution). The percentage of erythrocytes infected with a single parasite of reference lines of the ANKA strain of P. berghei consistently ranges between 0.5-2% at day 8 after infection, resulting in a mean multiplication rate of 10 per 24h. In addition, groups of mice (5) were infected with 20 to 200 or 200 to 2000 parasites of either ANKA wt-GFP or pbggcs 2 parasites (pbggcs 2 1 or pbggcs 2 2). Parasitemia (P = % of infected erythrocytes) was determined by microscopic examination of Diff Quick-stained thin smears of tail blood every day during a period of 10 days.

Analysis of the mosquito stages of pbggcs 2 parasites
Anopheles stephensi female mosquitoes were allowed to feed on mice infected with P. berghei ANKA wt-GFP and pbggcs 2 parasites (clones 1 or 2). Feeding was performed when the exflagellation rate of male gametocytes in the infected blood was between 1 and 2 per 10 fields [54]. In each experiment, mosquitoes were fed on one mouse and mosquitoes were kept at 21uC, feeding [55].
Ookinetes. Midguts (n = 25) filled with the blood bolus were dissected 24 hours p.i. and homogenized in 50 ml 16 PBS. From the homogenate, 5 ml were smeared and fixed on a slide by incubation in methanol for 10 minutes at 220uC. Parasites were counted on each slide after Giemsa staining and total numbers of ookinetes per midguts were calculated.
Oocysts. To determine the numbers of oocysts per mosquito, midguts (n$40) were dissected 2 days and 12 days p.i., fixed in 4% paraformaldehyde, and blocked with 3% bovine albumin serum. The blood bolus was removed from midguts dissected at day 2 p.i. and the ''empty'' midguts incubated with an antibody specific for the surface protein, Pbs21 (Mab 13.1) [56], of ookinetes and young oocysts. The antibody-stained oocysts at day 2 p.i. were detected and counted by fluorescent microscopy (4006 magnification) after incubation with a Rhodamine Red TM -X goat anti-mouse IgG (H+L) (Invitrogen). Oocysts from midguts dissected at day 12 p.i. were stained with 0.2% mercurochrome for 10 minutes at room temperature and counted using a light microscope (4006 magnification). In addition, midguts dissected at day 12 p.i. were stained with antibodies against two proteins, PbCap380 [22] and PbCS Mab 3D11 [23]. Antibody-stained oocysts were analyzed by fluorescence microscopy (4006 magnification) with Rhodamine Red TM -X goat anti-mouse IgG (H+L) or Texas RedH-X goat antirabbit IgG (H+L) (Invitrogen), respectively. The mean size of oocyst was determined at day 12 p.i. by measuring the diameter of 30 to 80 GFP-expressing oocysts from 5 to 8 individual mosquitoes using a fluorescence stereomicroscope Fluo Combi III (Leica, NL) and the Leica Application Suite AF software (Leica, NL).
To analyze the membrane potential of mitochondria of the oocyst stage, midguts dissected at day 12 p.i. were incubated with the 100 nM MitoTrackerH Red (Molecular Probes) for 10 minutes at room temperature. To stain mosquito midgut epithelial cells and establish the difference in staining pattern between oocysts and midgut epithelial cells, midguts were stained with 100 nM MitoTracker for a period of 30 minutes at room temperature. Stained oocysts or midguts were analyzed using a fluorescence microscope at a magnification of 4006 and 10006 oil immersion, excitation at 579 nm and emission at 600 nm.
Sporozoites. To analyze sporozoite production, mosquito salivary glands were dissected between 18 and 25 days p.i. and homogenized in 50 ml of 16 PBS. Salivary gland sporozoites were detected by light and fluorescence (GFP) microscopy. To analyze infectivity of sporozoites, infected mosquitoes (n = 50) were fed on naïve mice (2 per experiment) for a period of 15-20 minutes. The presence of parasites in the mice blood stream was monitored daily at day 6 to day 12 after feeding of the mosquitoes by analysis of Giemsa stained thin smears from tail blood.

Transmission electron microscopy analysis of oocysts
After infection of A. stephensi mosquitoes with ANKA wt-GFP and pbggcs 2 parasites, midguts were dissected at day 12, fixed in 2.5% glutaraldehyde (Electron Microscopy Sciences; EMS, Hatfield, PA) in 0.1 M sodium cacodylate buffer (pH 7.4) for 1 h at room temperature, and processed as described by [57] using a Philips 410 Electron Microscope (Eindhoven, the Netherlands) under 80 kV.

Statistical analysis
One way Anova (GraphPad prism 4.03, GraphPad Software, Inc. La Jolla, CA) or Mann Whitney U (STATVIEW 5.0 software, SAS Institute, Cary, NC) tests were used to determine statistical differences between parasite samples. In both cases, a P value of # 0.05 was used to establish statistical significance.