Biochemical and Genetic Analysis of the Phosphoethanolamine Methyltransferase of the Human Malaria Parasite Plasmodium falciparum*

The PfPMT enzyme of Plasmodium falciparum, the agent of severe human malaria, is a member of a large family of known and predicted phosphoethanolamine methyltransferases (PMTs) recently identified in plants, worms, and protozoa. Functional studies in P. falciparum revealed that PfPMT plays a critical role in the synthesis of phosphatidylcholine via a plant-like pathway involving serine decarboxylation and phosphoethanolamine methylation. Despite their important biological functions, PMT structures have not yet been solved, and nothing is known about which amino acids in these enzymes are critical for catalysis and binding to S-adenosyl-methionine and phosphoethanolamine substrates. Here we have performed a mutational analysis of PfPMT focused on 24 residues within and outside the predicted catalytic motif. The ability of PfPMT to complement the choline auxotrophy of a yeast mutant defective in phospholipid methylation enabled us to characterize the activity of the PfPMT mutants. Mutations in residues Asp-61, Gly-83 and Asp-128 dramatically altered PfPMT activity and its complementation of the yeast mutant. Our analyses identify the importance of these residues in PfPMT activity and set the stage for advanced structural understanding of this class of enzymes.

About 41% of the population of the world lives in malaria endemic areas, and every year more than 2 million people die from the disease (1). Malaria is caused by an intraerythrocytic protozoan parasite of the genus Plasmodium. Of the four species that infect humans, Plasmodium falciparum causes the most lethal form of the disease and has developed resistance to almost all the available drugs in the antimalarial armamentarium (2). New chemotherapeutic strategies are now needed to combat this disease. One strategy is to target the metabolic pathways that the parasite uses to synthesize new membranes, which are critical for parasite development and multiplication within erythrocytes (3). Various inhibitors of lipid metabolism have been shown to inhibit P. falciparum proliferation in vitro and in vivo, and several efforts are being made to advance these compounds for treatment of malaria.
Phosphatidylcholine (PtdCho) 3 composes half of the phospholipid content of the parasite membranes (4). Biochemical studies demonstrated that PtdCho synthesis occurs via two metabolic routes (5,6). The first route is the CDP-choline pathway, which uses host choline as a precursor. In this pathway choline is first phosphorylated to phosphocholine and then converted to CDP-choline. Subsequently, CDP-choline and diacylglycerol function as substrates for PtdCho synthesis. The second route is the serine decarboxylation-phosphoethanolamine methylation pathway (6). This pathway uses serine either transported from the host or generated by degradation of host proteins as a phospholipid precursor. The serine is first decarboxylated to produce ethanolamine, by an unknown serine decarboxylase. The ethanolamine is next phosphorylated by a parasite-specific ethanolamine kinase. A SAM-dependent triple methylation of the resulting phosphoethanolamine by a plant-like phosphoethanolamine methyltransferase results in the synthesis of phosphocholine (6). This product is then integrated into the CDP-choline pathway for the synthesis of phosphatidylcholine (6,7).
The transmethylation step of the serine decarboxylationphosphoethanolamine methylation pathway is catalyzed by a 266-amino acid phosphoethanolamine methyltransferase, PfPMT, that shares high homology with known and predicted phosphoethanolamine methyltransferases from plants and worms (6 -12). Biochemical studies in P. falciparum and genetic studies in yeast have demonstrated the specificity of the PfPMT enzyme for its substrate, phosphoethanolamine (6,7). Interestingly, no PfPMT homologs could be found in mammalian databases, suggesting that this protein could be an ideal target for development of novel antimalarial lipid inhibitors.
In this paper we report the biochemical and genetic characterization of PfPMT using yeast as a surrogate system. Scanning alanine mutagenesis of 24 conserved residues was performed followed by analysis of the activity of the mutant enzymes in vitro and in vivo using a functional complementation assay in yeast. Three residues, Asp-61, Gly-83, and Asp-128, were found to play a critical role in PfPMT catalysis and substrate binding.
Complementation Assay-pem1⌬pem2⌬ strains harboring wild-type or mutant Pfpmt CO under the regulatory control of the GAL1 promoter were grown overnight in SD medium supplemented with 1 mM choline, then washed three times and starved for choline by incubation in SD medium lacking choline (SDϪCho) overnight. Cells were washed again and inoculated at an A 660 ϭ 0.03 in SD or SG media containing choline, ethanolamine, or neither. The A 660 was measured at different time points during the 4 days of continuous culture.
Immunoblotting-S. cerevisiae wild-type and pem1⌬pem2⌬ cells harboring pYES2.1, pYES2.1 GAL1::PfPMT co , and mutant forms of Pfpmt co were grown in 30 ml of SD or SG media to A 600 ϭ 1 at 30°C. Y-PER reagent was used to extract the soluble protein fractions from the S. cerevisiae strains. Western blot analysis was performed using proteins from cells grown in minimal medium containing choline with or without glucose or galactose. Proteins were separated by SDS-PAGE, transferred to a nitrocellulose membrane, and analyzed by immunoblotting using anti-V5 primary antibody (Invitrogen, 1:5000 dilution) to detect Pfpmt and anti-hexokinase (1:10000) antibodies as a loading control.
Lipid Phosphorus Measurement-pem1⌬pem2⌬ cells harboring the pYES2.1 vector containing either wild-type PfPMT CO or mutant forms of PfPMT CO were pre-grown to mid-log phase at 30°C in synthetic uracilϪ dropout media containing 4% galactose (SG-URA medium) supplemented with ethanolamine (2 mM) and choline (2 mM). The cells were harvested by centrifugation, washed twice with water, and diluted to A 600 ϭ 0.03 in 50 ml of SG medium lacking uracil (SG-URA) but supplemented with ethanolamine (2 mM) and grown to A 600 ϳ 1.5. The cells were harvested by centrifugation and washed twice with water. The lipids were extracted as described previously (14). The phospholipids were separated by two-dimensional thin layer chromatography (TLC) on Silica 60 plates using chloroform/methanol/ammonium hydroxide (84.5/45/6.5 v/v/v) followed by chloroform/acetic acid/methanol/water (90/30/6/2.6 v/v/v). Lipids were visualized with iodine vapor and six classes of phospholipids (PtdCho, PtdEtn, PtdSer, PtdIns, Ptd 2 Gro (cardiolipin), phosphatidic acid) were excised from the plate and quantified by measuring phosphorus (15). The results are shown as the percentage of total lipid phosphorus in each phospholipid fraction. Data are the means Ϯ S.D. for three independent experiments.
Analysis by Circular Dichroism-The purified proteins were dialyzed against 5 mM phosphate buffer, pH 6.8, containing 100 mM sodium chloride, and just before the measurement, the buffer was exchanged with 5 mM phosphate buffer, pH 6.8, containing 100 mM sodium fluoride. The protein concentration was determined using the Edelhoch method (16). The circular dichroism spectrum of the purified recombinant protein (10 M) was recorded between 190 and 280 nm using a Jasco J-715 spectropolarimeter as previously described (17).
Enzyme Assays-The in vitro enzymatic activities of the purified wild-type and mutant Pfpmt proteins were determined by measuring the incorporation of radioactivity from [methyl-14 C]SAM into N-methylated derivatives of phosphoethanolamine as previously described (6). The kinetic properties of Pfpmt for the phosphoethanolamine substrate were determined under saturating concentrations of the cosubstrate SAM (2 mM) and with increasing concentrations of phosphoethanolamine (30 M to 2 mM). Likewise, the affinity of Pfpmt for its cosubstrate SAM was determined by using different concentrations of SAM (30 M to 2 mM) and a saturating concentration of phosphoethanolamine (2 mM). Fifty to 100 g of purified enzyme was used for the assay.

Mutagenesis and Functional Complementation in Yeast-
We have previously shown that a codon-optimized P. falciparum Pfpmt gene complements the choline auxotrophy of the yeast pem1⌬pem2⌬ mutant lacking the two phospholipid methyltransferases, Pem1p and Pem2p (7). To characterize residues in PfPMT that play a critical role in the transmethylation reaction leading to the formation of phosphocholine from phosphoethanolamine, a sequence alignment of PfPMT with PMTs from other species was analyzed (Fig. 1). Twenty-four conserved residues in or near the predicted catalytic motifs of PfPMT were mutated to alanine in the pYes-PfPMT CO vector, which harbors the codon-optimized PfPMT CO . In this vector expression of wild-type and mutant PfPMTs is under the regulatory control of the yeast GAL1 promoter that enables expression in the presence of galactose as the sole carbon source but is repressed when glucose is added. The resulting wildtype and mutated proteins harbor a C-terminal V5 and hexahistidine double tag, which allows monitoring of expression in yeast. Transformants were selected on cholinecontaining medium and then tested for growth on glucose or galactose media lacking choline and/or supplemented with ethanolamine. Of the 24 mutants tested, only three, Asp-61, Gly-83, and Asp-128, failed to complement the choline auxotrophy of pem1⌬pem2⌬ mutant (Fig. 2). The mutations in PfPMT in these three mutants reside in catalytic motifs I and P-I and outside of catalytic motif II, respectively (Fig. 1). As expected, transformants harboring an empty vector failed to grow in the absence of choline, whereas those harboring a wild-type copy of PfPMT grew normally on galactose medium lacking choline (Fig. 2). To further investigate the phenotype of these three mutants, pem1⌬pem2⌬ strains harboring Pfpmt D61A , Pfpmt G83A , Pfpmt D128A were inoculated at 10 4 cells/ml of culture, and cell growth was followed by measuring changes in cell density over time. As a control, pem1⌬pem2⌬ strains harboring an empty vector, wild-type PfPMT, or the mutated Pfpmt G63A and Pfpmt G153A , which were not affected by choline deprivation, were also tested. Using this assay, no growth could be detected in pem1⌬pem2⌬ harboring Pfpmt D61A , Pfpmt G83A , and Pfpmt D128A during 2 days of culture in galactose medium lacking choline. The phenotype of strains expressing Pfpmt G83A and Pfpmt D128A was indistinguishable from that of pem1⌬pem2⌬ harboring an empty vector (Fig. 3). By comparison, strains expressing Pfpmt D61A showed a severe growth defect during the first 48 h of incubation, but a significant increase in cell density was observed after 2.5 days of incubation. The pem1⌬pem2⌬ strains harboring wild-type PfPMT or mutated Pfpmt G63A or Pfpmt G153A exhibited a similar growth rate in galactose medium lacking or supplemented with choline (Fig. 3). To further investigate the side chain contribution in residues Asp-61 and Asp-128 in PfPMT activity, these amino acids were also mutated to glutamate and asparagine, and the growth of the pem1⌬pem2⌬ strains harboring Pfpmt D61E , Pfpmt D61N , Pfpmt D128E , and Pfpmt D128N was monitored. Similar to Pfpmt D61A and Pfpmt D128A mutants, Pfpmt D61E , Pfpmt D61N , Pfpmt D128E , and Pfpmt D128N mutants did not grow in galactose medium lacking choline (Fig. 3). As expected, none of the strains grew in glucose medium lacking choline. To rule out the possibility that the lack of activity was due to lack or reduced expression of recombinant proteins in yeast, immunoblot analyses were performed on protein extracts from pem1⌬pem2⌬ strains harboring wild-type or mutated PfPMT proteins using anti-V5 antibodies. Similar PfPMT expression levels were detected in all strains, and as expected, no signal could be detected using protein extracts from pem1⌬pem2⌬ strains harboring an empty vector (Fig. 4). As a control, antibodies against the yeast hexokinase were used and showed similar levels in all strains (Fig. 4).  I   I  I  I  F  I  T  O  M  I  I  F  I  T
Biochemical Characterization of Recombinant Pfpmt D61A , Pfpmt G83A , and Pfpmt D128A -To further examine the effect of the mutations on the folding and activity of the enzyme, recombinant wild-type (PfPMT) and mutant Pfpmt D61A , Pfpmt G83A , and Pfpmt D128A were expressed in E. coli, purified by affinity chromatography, and analyzed for folding properties by circular dichroism (CD) spectrophotometry (Fig. 7A). The CD profiles of Pfpmt D61A , Pfpmt G83A , and Pfpmt D128A were similar to that of wild-type PfPMT, with a strong peak around 222 nm indicative of extensive helix content. The molar ellipticity of each mutant is consistent with normally folded proteins when compared with wild type. The secondary structure content of Pfpmt G83A is comparable with wild type, whereasPfpmt D128A exhibited decreased helical content, and Pfpmt D61A exhibited increased helical content relative to wild type. Thermal unfolding experiments were consistent with the equilibrium measurements of secondary structure content; Pfpmt G83A exhibited a T m close to that of wild type, whereas Pfpmt D61A and Pfpmt D128A exhibited T m values above and below wild type, respectively (Fig. 7B). The thermal unfolding transition for Pfpmt D128A is broader than wild type, consistent with a less cooperative unfolding transition expected on the basis of lower helical content. Pfpmt D61A , Pfpmt D61E , Pfpmt D61N , Pfpmt G83A , Pfpmt D128A , Pfpmt D128E , and Pfpmt D128N activities were also measured in vitro and compared with that of wild-type PfPMT to determine the kinetic parameters of the SAM-dependent methylation of phosphoethanolamine. The apparent K m of wild-type PfPMT for SAM and phosphoethanolamine was estimated to be ϳ35.23 and 66.1 M, respectively, whereas its V max for SAM and phosphoethanolamine was 0.30 and 0.95 nmol/min/mg protein, respectively (supplemental Figs. S1 and S2). No activity could be detected with the mutated Pfpmt proteins.

DISCUSSION
The transmethylation of phosphoethanolamine by the malarial phosphoethanolamine methyltransferase, PfPMT, is critical for the synthesis of PtdCho in the human parasite P. falciparum (5,7). Genetic and pharmacological studies suggest that this step may also play an essential role during various stages of the parasite life cycle. 4 Understanding the biochemical and structural properties of this enzyme may help in the design of specific inhibitors that could block the enzyme activity and interfere with parasite proliferation inside human red blood cells.
PfPMT is a member of an unusual class of SAM-dependent methyltransferases also found in worms and plants but absent in mammals (8 -12). They share significant sequence similarity but seem to have evolved different structural and biochemical properties. The plant PMTs have two catalytic domains, with the N-terminal domain involved in the methylation of phosphoethanolamine into monomethylphosphoethanolamine and the C-terminal domains involved in the last two methylation reactions to form phosphocholine (8,10,11). Caenorhabditis elegans has two PMT enzymes (Pmt1 and Pmt2) with a length similar to that of the plant PMTs but each having a single catalytic domain located in either the N-terminal domain in Pmt1 or to the C-terminal domain in Pmt2 (9,12). Pmt1 catalyzes only the first methylation reaction, whereas Pmt2 catalyzes the last two methylation reactions. Although the malarial PfPMT shares homology with both plant and C. elegans PMTs, it is only half the size of these proteins and catalyzes the three methylation steps (6). All PMT enzymes have similar steady state kinetic properties for SAM and phosphoethanolamine. Although the function and biochemical properties of this class of enzymes have only started to be elucidated (5-12, 18, 19), nothing is known about their structure, and the residues within these enzymes that are critical for catalysis and substrate binding are unknown.
The studies reported here provide the first structure-function analysis of the activity of a PMT enzyme. These studies benefited from the previous genetic analysis in yeast, which showed that a codon-optimized PfPMT CO gene complements the choline auxotrophy of a mutant pem1⌬pem2⌬, lacking PEM1 and PEM2 genes encoding the methyltransferase enzymes essential for the transmethylation of PtdEtn to form PtdCho.
Twenty-four residues within and outside the catalytic core of the enzyme were mutated to alanine, and the corresponding genes were expressed in the pem1⌬pem2⌬ strain under the regulatory control of the GAL1 promoter. Our studies revealed that amino acids Asp-61, Gly-83, and Asp-128 play a critical role in enzyme activity. Asp-61 is located within the VLDIGS-GLG motif I of PfPMT. This amino acid is conserved in all PMTs. Its mutation to either asparagine or glutamate resulted in loss of complementation of pem1⌬pem2⌬ cells in the absence of choline and no activity in vitro. However, mutation of this residue to alanine resulted in a partial complementation of pem1⌬pem2⌬ growth in the absence of choline. Interestingly, phospholipid analysis of pem1⌬pem2⌬ cells harboring Pfpmt D61A revealed a PtdCho content of ϳ3%, which is similar to that obtained with the non-codon-optimized wild-type PfPMT (7). This amount of PtdCho has been shown to allow survival but not optimal growth of pem1⌬pem2⌬ cells in the absence of choline (7). Although this partial complementation suggests that the enzyme might have suboptimal activity, in vitro transmethylation activity assay using recombinant Pfpmt D61A enzyme showed no activity. This discrepancy could be due to the differences in sensitivity between the genetic survival assay in yeast and the in vitro transmethylation assay.
Gly-83 is located within the motif P-I HGID of Pfpmt. This residue is also conserved in all PMTs, and like Asp-61, the mutation to alanine resulted in enzymatic loss of affinity for SAM and phosphoethanolamine as well as a concomitant complete loss of function in vivo. Consistent with this loss of activity, phospholipid analysis of pem1⌬pem2⌬ cells harboring this mutant enzyme revealed a PtdCho content that was reduced by 96% compared with that of pem1⌬pem2⌬ cells harboring a wild-type PfPMT.
Asp-128 is located outside the predicted motif II NNFDLIYS. This residue is also conserved in all PMTs and is strictly essential for PfPMT activity. Its mutation to alanine resulted in a complete loss of activity in vitro and lack of complementation in yeast. Consistent with this loss of activity, phospholipid analysis of pem1⌬pem2⌬ cells harboring this mutant enzyme revealed a PtdCho content of ϳ2%.
The molar ellipticity of the mutants relative to WT PfPMT makes clear that the loss of function of the mutants does not result from failure to fold. The secondary structure content of D128A and D61A, which is lower and higher than WT, respectively, is consistent with the lower and higher thermal stabilities. Yet both display complete loss of enzymatic activity. We hypothesize that the side chain of Asp-128 is involved in favorable electrostatic interactions, whereas that for Asp-61 is involved in unfavorable electrostatic interactions, thus leading to the observed decrease (D128A) or increase (D128A) in stability upon mutation to alanine. NMR structural studies with the wild-type PfPMT and point mutants are under way. The ionization states and pK a values of the critical aspartate residues, which can be determined by NMR, will be especially valuable for elucidating the physical basis of catalysis and binding. The mechanism underlying the loss-of-function in G83A is intriguing. Glycine is unlikely to play a direct role in catalysis or binding and yet the mutation does not appear to substantially alter the structure or stability. A potential mechanism is the loss of flexibility needed to undergo required conformational change. Such a dynamic effect could be revealed by thermodynamic or NMR relaxation studies.
In conclusion, we have identified three residues in PfPMT that are critical for enzymatic activity. Further investigations are planned to elucidate the structural, thermodynamic, and catalytic roles played by these residues. The insights derived from these studies will enable intelligent screening of potential inhibitors, whether by NMR, high throughput screening methods, or in silico.