Phosphoglycolate phosphatase is a metabolic proofreading enzyme essential for cellular function in Plasmodium berghei

Plasmodium falciparum (Pf) 4-nitrophenylphosphatase has been shown previously to be involved in vitamin B1 metabolism. Here, conducting a BLASTp search, we found that 4-nitrophenylphosphatase from Pf has significant homology with phosphoglycolate phosphatase (PGP) from mouse, human, and yeast, prompting us to reinvestigate the biochemical properties of the Plasmodium enzyme. Because the recombinant PfPGP enzyme is insoluble, we performed an extended substrate screen and extensive biochemical characterization of the recombinantly expressed and purified homolog from Plasmodium berghei (Pb), leading to the identification of 2-phosphoglycolate and 2-phospho-L-lactate as the relevant physiological substrates of PbPGP. 2-Phosphoglycolate is generated during repair of damaged DNA ends, 2-phospho-L-lactate is a product of pyruvate kinase side reaction, and both potently inhibit two key glycolytic enzymes, triosephosphate isomerase and phosphofructokinase. Hence, PGP-mediated clearance of these toxic metabolites is vital for cell survival and functioning. Our results differ significantly from those in a previous study, wherein the PfPGP enzyme has been inferred to act on 2-phospho-D-lactate and not on the L isomer. Apart from resolving the substrate specificity conflict through direct in vitro enzyme assays, we conducted PGP gene knockout studies in P. berghei, confirming that this conserved metabolic proofreading enzyme is essential in Plasmodium. In summary, our findings establish PbPGP as an essential enzyme for normal physiological function in P. berghei and suggest that drugs that specifically inhibit Plasmodium PGP may hold promise for use in anti-malarial therapies.

The haloacid dehalogenase superfamily (HADSF) 5 is a large family of enzymes consisting mainly of phosphatases and phosphotransferases, which are both intracellular and extracellular in nature. These enzymes are characterized by the presence of a core Rossmanoid fold and a cap domain (1,2). Studies of HADSF members have focused on identifying their physiological substrates by screening a wide range of metabolites that include sugar phosphates, lipid phosphates, nucleotides, as well as phosphorylated amino acids and co-factors. This approach has helped us understand the physiological relevance of these enzymes in various cellular processes, such as cell wall synthesis, catabolic and anabolic pathways, salvage pathways, signaling pathways, and detoxification (3)(4)(5)(6)(7)(8)(9)(10)(11)(12)(13). Apart from dephosphorylating metabolites, HADSF members have also been known to dephosphorylate proteins, and such members are characterized by the absence of the cap domain (1,2). A largescale study reported by Huang et al. (5) has identified a HADSF member from Salmonella enterica that catalyzes dephosphorylation of more than 100 phosphorylated substrates (5). This extended substrate specificity is a common observation in HADSF members and often leads to a confounding situation where determining the physiological substrate of such promiscuous enzymes becomes a challenging task.
Recent studies have identified and characterized HADSF members from the apicomplexan parasite Plasmodium (4,10,(13)(14)(15)(16). HADSF members from Plasmodium have been found to be involved in processes that lead to the development of resistance to the drug fosmidomycin, which inhibits isoprenoid biosynthesis (4). Also, these enzymes show considerable activity toward nucleotide monophosphates, phosphorylated co-factors, and generic substrates such as p-nitrophenylphosphate (pNPP) and ␤-glycerophosphate. A HADSF member that was annotated as 4-nitrophenylphosphatase from Plasmodium falciparum (gene ID PF3D7_0715000) was characterized by Knöckel et al. (15) and proposed to be involved in dephosphor-ylation of thiamine monophosphate, the precursor of the active form of vitamin B1 (thiamine pyrophosphate). In vitro assays of the purified recombinant enzyme showed that this protein displayed similar specific activities toward thiamine monophosphate and other substrates (ADP, ATP, CTP, Glc-6-P, Fru-6-P, and pyridoxal phosphate) (15). An independent BLASTp search conducted by us revealed that this protein sequence has significant homology (28 -30%) with phosphoglycolate phosphatase (PGP) from yeast, human and mouse (Fig. 1). The His 6tagged recombinant P. falciparum (Pf) 4-nitrophenylphosphatase, when expressed in Escherichia coli, was found to be completely insoluble. However, Plasmodium berghei (Pb) 4-nitrophenylphosphatase (gene ID PBANKA_1421300) (referred to as PbPGP hereafter), which shares 69.6% identity (Fig. 1B) with its Pf homolog, expressed in the soluble form in E. coli and could be purified to homogeneity. Here we report the biochemical characterization and essentiality of PbPGP. An extended substrate screen identified 2-phosphoglycolate and 2-phospho-L-lactate as relevant physiological substrates in addition to the generic substrates pNPP and ␤-glycerophosphate. Attempts at gene ablation showed that the PbPGP gene cannot be disrupted in P. berghei despite the loci being nonrefractory for genetic recombination. Our findings emphasize the importance of the "metabolic proofreading" process, which involves clearance or modification of toxic cellular metabolites generated as a consequence of error in substrate recognition by enzymes of intermediary metabolism. This process is universal and analogous to the DNA proofreading observed in polymerases and

P. berghei phosphoglycolate phosphatase
aminoacyl-tRNA synthetases (17). Our studies of PbPGP establish the essential physiological nature and biochemical function of this conserved cytosolic enzyme and suggest that drugs that specifically inhibit parasite phosphoglycolate phosphatase can be promising anti-malarial agents.

Biochemical characterization of recombinant PbPGP
BLASTp analysis of Pf and PbPGP protein sequences showed 28 -30% sequence homology with phosphoglycolate phosphatase, a conserved protein, present across eukaryotes from yeast to mouse, including humans, involved in metabolic proofreading (Fig. 1A). The Pf and Pb protein sequences show 69.6% identity (Fig. 1B).
Upon expression of C-terminal His 6 -tagged PfPGP in the Rosetta DE3 pLysS strain of E. coli, the protein was found to be present completely in the insoluble fraction (Fig. S1A). This was unlike the Strep-tagged PfPGP, which was reported to be present in small quantities in the soluble fraction and, hence, amenable to purification. Therefore, we made use of the protein solubility prediction software PROSOII and found that homologs of PfPGP from other Plasmodium species were predicted to be soluble (Fig. S1B). Hence, the previously uncharacterized P. berghei homolog was chosen for further biochemical studies and physiological investigations. PbPGP was expressed in the E. coli strain Rosetta DE3 pLysS and purified to homogeneity by Ni-NTA affinity chromatography (Fig. S1C), followed by sizeexclusion chromatography ( Fig. 2A).
PbPGP on analytical gel filtration using a Sephacryl S-200 column showed a mass of about 78 kDa, whereas the theoretical mass is 37 kDa, indicating that the protein is a dimer (Fig. 2, B and C). When further analyzed in the presence of 1 M NaCl, there was a shift in oligomeric state of the protein from dimer toward monomer, suggesting that the oligomers are held by electrostatic interactions (Fig. 2, B and C).

P. berghei phosphoglycolate phosphatase
very high activity on 2-phosphoglycolate and 2-phospho-L-lactate in addition to the generic substrates pNPP and ␤-glycerophosphate (Fig. 2D). It should be noted that the enzyme was stereospecific for 2-phospho-L-lactate and showed no activity on 2-phospho-D-lactate.

Kinetic studies of PbPGP
PbPGP showed maximum activity at pH 7.0 and preferred Mg 2ϩ as a co-factor over other divalent cations (Fig. 3, B and C). The substrate saturation plots for ␤-glycerophosphate, 2-phosphoglycolate, and 2-phospho-L-lactate were hyperbolic (Fig. 3, D-F) and were fit to the Michaelis-Menten equation to obtain kinetic parameters such as K m and V max (Table 1). PbPGP has a higher K m value for 2-phosphoglycolate (3.3-and 11.4-fold) and 2-phospho-L-lactate (27.4-and 6.4-fold) compared with that of murine PGP and yeast Pho13. The k cat value for PbPGP for 2-phosphoglycolate is 11.4-and 3.9-fold higher and for 2-phospho-L-lactate is 37-and 8.9-fold higher compared with that of the murine and yeast homologs, respectively. The catalytic efficiency (k cat /K m ) for 2-phosphoglycolate was 3.5-fold higher and 2.9-fold lower compared with its murine and yeast homologs, respectively. With 2-phospho-L-lactate as substrate, the parasite enzyme has a similar catalytic efficiency as its murine and yeast homologs.

Probing the essentiality of PbPGP and localization in P. berghei
The pJAZZ linear knockout vector for PbPGP was generated by following the strategy described by Pfander et al. (18). Drugresistant parasites were not obtained in the first transfection attempt. In the second attempt, although drug-resistant parasites were obtained, genotyping by PCR revealed nonspecific integration of the marker cassette. These parasites were positive by PCR for both the PbPGP gene and the human dihydrofolate reductase (hDHFR) marker but were negative for specific 5Ј and 3Ј integration PCRs (Fig. S4). Because it was not possible to obtain knockout parasites, a conditional knockdown (at the protein level) strategy was employed by tagging the gene for PbPGP with a regulatable fluorescent affinity (RFA) tag, where the stability of the fusion protein is conditional to the binding of the small molecule trimethoprim. The conditional knockdown vector was also generated by following the recombineering strategy and validated by PCR (Fig. 4). Transgenic parasites were obtained in the first transfection attempt, and genotyping by PCR showed the presence of a single homogenous population with correct insertion of the RFA tag (Fig. 4F). Nevertheless, it was observed that the reduction in the levels of RFAtagged protein upon removal of TMP varied between 30 -60% across experiments, and complete knockdown could not be achieved (Fig. 5, A-C). As a consequence, there was no significant difference in growth rate between parasites grown in mice fed with or without trimethoprim (Fig. 5, D and E). The transgenic RFA-tagged P. berghei parasites were employed to determine localization of PbPGP, and upon microscopic observation, a cytosolic GFP signal was observed in all intraerythrocytic stages (Fig. 5F). rum. The authors proposed a novel role for this HADSF member and suggested involvement in vitamin B1 homeostasis. We found the P. falciparum 4-nitrophenylphosphatase sequence to have homology with human, mouse, and yeast phosphoglycolate phosphatases. An extended substrate specificity screen of the recombinant P. berghei enzyme revealed that, indeed, this protein is phosphoglycolate phosphatase, which is mainly involved in detoxification, having very high activity on 2-phosphoglycolate and 2-phospho-L-lactate with no activity on thiamine monophosphate. 2-Phosphoglycolate has been reported to be formed during repair of free radical-mediated damage of DNA ends (19), and accumulation of this metabolite in the cell Table 1 Kinetic parameters of P. berghei PGP compared with that of homologs from yeast and mouse Data represent mean Ϯ S.E. (n ϭ 2).

P. berghei phosphoglycolate phosphatase
leads to inhibition of the key glycolytic enzyme triosephosphate isomerase (Fig. 6). Studies of phosphoglycolic acid phosphatases from yeast and mouse have demonstrated that this enzyme also performs metabolic proofreading by catabolizing the substrates 2-phospho-L-lactate and 4-phosphoerythronate, which are products of enzymatic side reactions. Activity of PbPGP on 4-phosphoerythronate could not be tested because of nonavailability of the compound. 2-Phospho-L-lactate, generated by phosphorylation of L-lactate by pyruvate kinase, is known to inhibit phosphofructokinase and 4-phosphoerythronate, which is a product of GAPDH side reaction, is known to inhibit 6-phosphogluconate dehydrogenase (Fig. 6) (20). Because of the detrimental effect of these metabolites, it becomes essential to clear the cell of these metabolic toxins. This is reflected by the fact that phosphoglycolate phosphatase is an essential gene in the mouse (21). Also, in Arabidopsis, knockout of PGLP1 isoform leads to impaired post-germination development of primary leaves (22). Plasmodium, in its intraerythrocytic stages, experiences very high levels of oxidative stress (23), leading to increased reactive oxygen species (ROS) production, which can damage its DNA, the repair of which will result in generation and accumulation of 2-phosphoglycolate. The parasite performs lactic acid fermentation and secretes large amounts of lactate into the medium, most of which is L-lactate (93-94%) in addition to a small proportion of D-lactate (6 -7%), which is known to be produced through the methylglyoxal pathway (24). This lactate can accumulate and be phosphorylated in the cell to give rise to 2-phospholactate. In a recent study, Dumont et al. (16), by metabolite profiling of WT and ⌬pfpgp P. falciparum, concluded that PfPGP has specificity for 2-phospho-D-lactate, whereas our study contradicts this inference and provides direct evidence for the sole substrate specificity for the L-isomer. In their experiment, ⌬pfpgp parasites, when grown under normal culture conditions or in the presence of 2 mM L-lactate, showed a similar 12-fold higher accumulation of phospholactate compared with WT parasites grown under the same culture conditions. However, in the presence of increasing concentrations of D-lactate in the culture medium, they observed a dose-dependent increase in the accumulation of phospholactate that was not significantly different between WT and ⌬pfpgp parasites. Although this rules out the absence of PfPGP activity being the cause for 2-phospho-D-lactate accumulation, the authors have concluded that PfPGP utilizes 2-phospho-Dlactate as a substrate (16). The physiological reasons for these observations of Dumont et al. (16) can be rationalized in the following manner. As L-lactate is the predominant isomer produced in high concentrations in the cell, externally added L-lactate may not be taken up inside the cell or, even when taken up, might not significantly perturb the intracellular L-lactate concentration. Therefore, in the experiment of Dumont et al. (16), addition of L-lactate to the culture medium did not lead to an increase in levels of phospholactate in WT or ⌬pfpgp parasites. D-lactate is produced in the parasite at very low levels, and exogenously added D-lactate might be acted on by pyruvate kinase to form 2-phospho-D-lactate. This can happen in both

P. berghei phosphoglycolate phosphatase
WT and ⌬pfpgp parasites and the absence of a significant difference in the levels of phospholactate accumulation between WT and ⌬pfpgp parasites when grown in the presence of D-lactate (16) is expected, as our studies show that PGP is specific for only 2-phospho-L-lactate. Further, Dumont et al. (16) performed metabolite profiling of WT and ⌬glo1 (impaired in D-lactate production) parasites in the presence of methyl glyoxal in the culture medium and observed similar levels of phospholactate accumulation in both parasites. The authors justify this observation by speculating that either methly glyoxal is converted to D-lactate in erythrocytes and then transported to the parasite or directly converted to D-lactate in the parasite by involvement of the apicoplast glyoxalase-1 (16). Either way, D-lactate levels in the parasite increase, and it is phosphorylated and accumulates in both WT and ⌬glo1 parasites as phospholactate, in spite of the presence of the PGP gene. This again goes to show that PGP does not act on the D-isomer. As pyruvate kinase is known to have a higher binding affinity for D-lactate compared with that for L-lactate (25), accumulation of phospholactate in WT and ⌬pfpgp parasites grown on D-lactate, and parasites grown on methyl glyoxal could be a consequence of preferential activity of pyruvate kinase on D-lactate. The indirect inferences provided by Dumont et al. (16) are akin to a "phenocopy" witnessed as a consequence of pyruvate kinase activity on D-lactate rather than a true phenotype associated with phosphoglycolate phosphatase deficiency. Our results regarding the purified enzyme directly show that PbPGP acts only on 2-phospho-L-lactate and not on 2-phospho-D-lactate (Fig. 2D). We further validated this by performing enzyme assays with 1 mM 2-phospho-L-lactate in the presence or absence of 10 mM 2-phospho-D-lactate. The absence of a significant change in specific activity clearly shows that 2-phospho-D-lactate does not bind to the enzyme (Fig. 3A). This observation is consistent with that of the murine homolog of PbPGP, which also acts only on 2-phospho-L-lactate (20). In addition to the evidence above, the possible difference in substrate specificity across the enzymes from the two Plasmodium species, P. falciparum and P. berghei, was also addressed by taking recourse to sequence and structural analysis of the proteins. Both proteins are highly identical, and the residues around the four HAD motifs are highly conserved (Fig.  S5A). Both protein sequences were subjected to homology modeling, and both modeled structures aligned without any gross structural differences (Fig. S5, B and C). This strongly suggests that PfPGP, like PbPGP, would also have specificity for only 2-phospho-L-lactate.
The recombinant human pyruvate kinase M2 isoform has been shown to phosphorylate L-lactate, leading to the production of 2-phospho-L-lactate, which, in turn, has been shown to inhibit phosphofructokinase-2 activity in crude lysates of HCT116 cells and activity of the recombinant phosphofructokinase-fructose 1,6-bisphosphatase isozymes PFKFB3 and PFKFB4 (20). Interestingly, in yeast, knockout of PHO13 (a PGP homolog) is viable, as yeast performs alcohol fermentation instead of lactate fermentation and, hence, does not accumulate phospholactate. Also, inhibition of the pentose phosphate pathway caused by accumulation of 4-phospho D-erythronate is countered by transcriptional up-regulation of pentose phos-phate pathway enzymes (20). Plasmodium has two genes coding for phosphofructokinase, one on chromosome 9 (PfPFK9) and the other on chromosome 11 (PfPFK11), and only PfPFK9 has been shown to be functional. It has been reported that, unlike the host enzyme, PfPFK9 lacks regulation by fructose 1,6-bisphosphate, phosphoenolpyruvate, and citrate (26). In such a scenario, we speculate that 2-phospho-L-lactate might directly inhibit PfPFK9 to regulate glycolysis, as knockout of PbPGP is not possible. This selective mode of regulation might be unique to Plasmodium.
In Plasmodium, where glycolysis is the sole source of ATP in asexual stages (27), the parasite cannot afford inhibition of its critical enzymes, such as PFK and triosephosphate isomerase, arising from accumulation of toxic metabolites. Hence, having a metabolic proofreading/detoxifying enzyme becomes vital for its survival. An inability to obtain knockout parasites indicates the essentiality of this protein for parasite survival during asexual stages. To rule out the possibility of the loci being refractory for genetic recombination, RFA tagging was attempted, and a homogenous population of transfectants with the RFA tag integrated at the right loci was obtained. Having established that the locus is amenable for genetic manipulation, a conditional knockdown strategy at the protein level was adopted, making use of the RFA tag (28). Conditional knockdown at the protein level showed only a 30 -60% reduction, and, hence, the parasites were viable (Fig. 5, A-E). A similar observation has been described for yoelipain, where the authors were neither able to knockout nor achieve significant knockdown of protein levels to see a growth difference. Therefore, it was concluded that the gene is essential during intraerythrocytic stages (29). Our results are similar and indicate the essentiality of PbPGP in asexual stages. This conclusion regarding the gene essentiality of PbPGP is in agreement with the findings of Dumont et al. (16) regarding PfPGP, where ⌬pfpgp parasites show a growth defect. Further biochemical and structural studies of PGP could pave the way for the rational design of inhibitors with potent anti-malarial activity.

Materials
All chemicals, molecular biology reagents, and medium components were from Sigma-Aldrich, New England Biolabs, Gibco, Invitrogen, US Biochemicals, Spectrochem, and Himedia. The E. coli strain XL-1 blue, the expression strain Rosetta (DE3) pLysS, and plasmids pET22b and pET23d were from Novagen. The pJAZZ library clone (PbG02_B-53b06); plasmids pSC101BAD, R6K Zeo/pheS, and GW_R6K_GFP-mut3; and E. coli pir strains were procured from PlasmoGem (Sanger Institute). E. coli TSA cells were from Lucigen. Plasmid pGDB was a kind gift from Dr. Vasant Muralidharan (University of Georgia). The P. falciparum 3D7 strain and P. berghei ANKA strain were procured from MR4. The Amaxa 4D Nucleofector and P5 Nucleofection kits were from Lonza. Gene sequencing of the various plasmid constructs was by the Sanger sequencing method. Sequences of oligonucleotides used are provided in Table S1.

P. berghei phosphoglycolate phosphatase Bioinformatics analysis
The PfPGP (PlasmoDB gene ID PF3D7_0715000) protein sequence obtained from the PlasmoDB database was subjected to a homology search against the nonredundant database at the NCBI using the BLASTp algorithm. Clustal Omega (30) was used to generate multiple sequence alignment. ProsoII (31) was employed to predict the solubility of proteins upon heterologous expression in the E. coli system.

Cloning expression and purification of PfPGP and PbPGP
All expression plasmids with the desired gene of interest were generated in the XL-1 blue strain of E. coli cells. Table S1 lists the oligonucleotide sequences used. The PfPGP gene was amplified by PCR from P. falciparum genomic DNA using the primers P1 and P2. The purified PCR product was digested with the restriction enzymes NcoI and XhoI and ligated with the double-digested plasmid pET23d. Chemically competent E. coli XL-1 blue cells were transformed with the ligation mixture, and transformants were selected on Luria-Bertani medium plates containing ampicillin (100 g ml Ϫ1 ) and tetracycline (10 g ml Ϫ1 ). The clones were validated by DNA sequencing. A plasmid isolated from the confirmed clone (pET23d_PfPGP) was used to transform the Rosetta DE3 pLysS strain of E. coli.
A single colony of transformed Rosetta DE3 pLysS was inoculated into 10 ml of terrific broth (TB) medium containing ampicillin (100 g ml Ϫ1 ) and chloramphenicol (34 g ml Ϫ1 ) and incubated at 37°C at 180 rpm for 12-15 h. The overnight culture was inoculated (1 ml/100 ml) into 800 ml of TB medium containing ampicillin and chloramphenicol. The cells were allowed to grow at 37°C and 180 rpm till 0.6 A 600 was reached and thereafter induced with isopropyl 1-thio-␤-D-galactopyranoside at a final concentration of 0.3 mM. Induction was carried out at 18°C and 180 rpm for 15 h. The cells were harvested by centrifugation at 4000 ϫ g for 15 min at 4°C. The harvested cells were resuspended in 25 ml of lysis buffer (20 mM Tris HCl (pH 7.4), 100 mM NaCl, 1 mM DTT, 0.1 mM PMSF, and 10% w/v glycerol), and the cell suspension was lysed by passing through a French pressure cell at 1000 p.s.i. for 7 cycles. The lysate was centrifuged at 18,000 ϫ g for 30 min at 4°C. The supernatant was collected, mixed with a 0.5-ml packed volume of Ni-NTAagarose beads, and incubated at 4°C for 3 h with rotation. The suspension was then transferred to a glass column, and the flow-through was collected. The beads were washed using lysis buffer containing increasing concentrations of imidazole. The pellets from the centrifuged cell lysate, unbound (flowthrough), and wash fractions were analyzed by 12% SDS-PAGE and gel-stained with Coomassie Brilliant Blue R-250.
PbPGP was amplified by PCR from P. berghei genomic DNA using the primers P3 and P4. The PCR product was cloned into pET22b between sites NdeI and XhoI. The cloning procedure and protocol for protein expression were similar to that of PfPGP. The fractions from Ni-NTA chromatography containing PbPGP were concentrated and subjected to further purification by size-exclusion chromatography on a Sephacryl S-200 column (1.5 ϫ 60 cm).

Determination of the oligomeric state
The oligomeric state of PbPGP was determined by analytical size-exclusion chromatography using a Sephacryl S-200 (1 ϫ 30 cm) column attached to an AKTA Basic HPLC system. The column was equilibrated using 100 mM Tris HCl (pH 7.4) and 100 mM KCl at a 0.8 ml min Ϫ1 flow rate and calibrated using the molecular mass standards: ␤-amylase (200 kDa), alcohol dehydrogenase (150 kDa), BSA (66 kDa), carbonic anhydrase (29 kDa), and cytochrome c (12.4 kDa). 100 l of PbPGP at 1 mg ml Ϫ1 concentration was injected into the column and eluted with equilibration buffer with monitoring at 280 nm. The molecular mass of PbPGP was estimated by interpolating the elution volume on a plot of logarithm of molecular mass standards on the y axis and elution volume on the x axis. Gel filtration was performed with and without NaCl in the equilibration buffer.

Synthesis of 2-phospholactate
Synthesis of both D and L phospholactate was carried out following an available procedure (20). The details of the protocol and characterization of the molecules are provided in Figs. S2 and S3.

Enzyme assays
A comprehensive substrate screen comprising various classes of molecules, such as nucleoside phosphates, sugar phosphates, co-enzymes, amino acid phosphates, etc., was performed. The assay was carried out in 100 mM Tris HCl (pH 7.4), 2 mM substrate, and 1 mM MgCl 2 in a volume of 100 l. The reaction mixture was preincubated at 37°C for 1 min, the assay was initiated using 2 g of enzyme, and the reaction was allowed to proceed at 37°C for 5 min. The reaction was stopped by addition of 20 l of 70% TCA, and 1 ml of freshly prepared Chen's reagent (water, 6 N sulfuric acid, 2.5% ammonium molybdate, and 10% L-ascorbic acid mixed in the ratio 2:1:1:1) was added, mixed thoroughly, and incubated at 37°C for 1.5 h. The color developed was measured against a blank (reaction mixture to which enzyme was added after addition of TCA) at 820 nm. Specific activity was calculated using the ⑀ value of 25,000 M Ϫ1 cm Ϫ1 .
The pH optimum of PbPGP was determined by performing the assay in a mixed buffer containing 50 mM each of glycine, MES, Tris at a different pH level, 1 mM MgCl 2 , and 1 mM pNPP as substrate in a 100-l volume. The reaction mixture was preincubated at 37°C for 1 min; the assay was initiated using 0.2 g of enzyme, and the reaction was allowed to proceed at 37°C for 2 min, stopped using TCA, and processed using Chen's reagent as described above.
The preferred divalent metal ion was identified using 10 mM pNPP as substrate and different salts such as MgCl 2 , MnCl 2 , CaCl 2 , CuCl 2 , and CoCl 2 at a final concentration of 1 mM in a 250-l reaction mixture containing 50 mM Tris HCl (pH 8). The reaction was initiated with 0.26 g of enzyme, and conversion of pNPP to p-nitrophenol was continuously monitored at 405 nm at 37°C. The slope of the initial 20 s of the progress curve was used to calculate specific activity using an ⑀ value of 18,000 M Ϫ1 cm Ϫ1 .

P. berghei phosphoglycolate phosphatase
Kinetic studies K m values for 2-phosphoglycolate, 2-phospho-L-lactate, and ␤-glycerophosphate was determined by measuring the initial velocity at varying substrate concentrations ranging from 0.5-15 mM for 2-phosphoglycolate and 2-phospho-L-lactate and 0.25-30 mM for ␤-glycerophosphate. The concentration of MgCl 2 was fixed at 5 mM, with the reaction buffer being 200 mM Tricine-NaOH (pH 7.4). The reaction, in a volume of 100 l, was initiated with 1.89 g of enzyme, allowed to proceed at 37°C for 2 min, stopped using TCA, and processed using Chen's reagent as described above. Specific activity was plotted as a function of substrate concentration, and the data points were fitted to the Michaelis-Menten equation using GraphPad Prism V5 to determine the kinetic parameters (32).

Generation of P. berghei transfection vectors
The library clone for P. berghei PGP (PbG02_B-53b06) was obtained from PlasmoGem. The procedure for knockout and tagging construct generation has been described previously (33,18) and is provided in detail in the supporting information.

Cultivation and transfection of P. berghei
Male/female BALB/c mice aged 6 -8 weeks were used for cultivation and transfection of P. berghei. Glycerol stock of WT P. berghei ANKA parasites was injected into a healthy male BALB/c mouse. Parasitemia was monitored by microscopic observation of Giemsa-stained smears of blood drawn from a tail snip. Transfection of the parasites was done by following the protocol described by Janse et al. (34), using Amaxa 4D Nucleofector (P5 solution and FP167 program) followed by injection into two mice. For PbPGP knockout, drug-resistant parasites were selected by feeding infected mice with pyrimethamine in drinking water (7 mg in 100 ml), whereas parasites with the PbPGP RFA tag were selected by feeding infected mice with trimethoprim in drinking water (30 mg in 100 ml). Drug-resistant parasites were harvested in heparin solution (200 units ml Ϫ1 ) made in RPMI 1640 medium. Glycerol stocks were made by mixing 300 l of 30% glycerol and 200 l of the harvested blood and stored in liquid nitrogen. Validation of the drugresistant parasites was done by PCR.

Conditional knockdown of PbPGP in P. berghei
Glycerol stock of PbPGP RFA-tagged parasites was injected into a healthy BALB/c mouse. Parasitemia was monitored by microscopic observation of Giemsa-stained smears of blood drawn from a tail snip. Trimethoprim pressure was maintained throughout the growth period. Upon parasitemia reaching 5-10%, about 500 l of infected blood was collected in 500 l of RPMI 1640 solution containing heparin. 100 l of this parasite containing suspension was injected into a fresh mouse that was fed with trimethoprim in drinking water and a second 100 l to another mouse that was not fed trimethoprim. Parasites were harvested from both mice after 6 days and subjected to Western blotting. The entire experiment was repeated twice.
Glycerol stock of PbPGP RFA-tagged P. berghei was injected into a healthy BALB/c mouse, and parasitemia was monitored by Giemsa-stained smears. Blood was harvested in heparin solution when parasitemia reached 0.5-1%, and 1.7 ϫ 10 5 parasites were injected into two batches of mice (n ϭ 5). One batch was fed with drinking water containing trimethoprim (30 mg in 100 ml) and the other with water lacking trimethoprim. Parasitemia was monitored regularly starting from day 2 post-injection by counting parasites in Giemsa-stained smears. The growth rate was determined by plotting the percentage of parasitemia on the y axis against time (number of days) on the x axis. The mortality rate among infected mice fed trimethoprimcontaining water or not was also determined by plotting the percentage of survival of mice on the y axis against time (number of days) on the x axis.

Localization of PGP in P. berghei
PbPGP RFA-tagged parasites were harvested in heparin solution and centrifuged at 2100 ϫ g for 5 min, and the supernatant was discarded. The cells were resuspended in 1ϫ PBS containing Hoechst 33342 (10 g ml Ϫ1 ) and incubated at room temperature for 15 min. Thereafter, the cells were collected, washed once with 1ϫ PBS, resuspended in 70% glycerol, and dispersed on poly-L-lysinecoated coverslips that were mounted on glass slides, sealed, and stored at 4°C. The slides were observed under an oil immersion objective (ϫ100) of a Zeiss LSM 510 Meta confocal microscope.

Ethics statement
Animal experiments involving handling of BALB/c mice were performed by adhering to standard procedures prescribed by the Committee for the Purpose of Control and Supervision of Experiments on Animals (CPCSEA), a statutory body under the Prevention of Cruelty to Animals Act of 1960, and the Breeding and Experimentation Rules of 1998, Constitution of India. This study (Project HB006/201/CPCSEA) was approved by the institutional animal ethics committee of the Jawaharlal Nehru Centre for Advanced Scientific Research, which is under the purview of CPCSEA.