Carotenoid Biosynthesis in Intraerythrocytic Stages of Plasmodium falciparum*

Carotenoids are widespread lipophilic pigments synthesized by all photosynthetic organisms and some nonphotosynthetic fungi and bacteria. All carotenoids are derived from the C40 isoprenoid precursor geranylgeranyl pyrophosphate, and their chemical and physical properties are associated with light absorption, free radical scavenging, and antioxidant activity. Carotenoids are generally synthesized in well defined subcellular organelles, the plastids, which are also present in the phylum Apicomplexa, which comprises a number of important human parasites, such as Plasmodium and Toxoplasma. Recently, it was demonstrated that Toxoplasma gondii synthesizes abscisic acid. We therefore asked if Plasmodium falciparum is also capable of synthesizing carotenoids. Herein, biochemical findings demonstrated the presence of carotenoid biosynthesis in the intraerythrocytic stages of the apicomplexan parasite P. falciparum. Using metabolic labeling with radioisotopes, in vitro inhibition tests with norflurazon, a specific inhibitor of plant carotenoid biosynthesis, the results showed that intraerythrocytic stages of P. falciparum synthesize carotenoid compounds. A plasmodial enzyme that presented phytoene synthase activity was also identified and characterized. These findings not only contribute to the current understanding of P. falciparum evolution but shed light on a pathway that could serve as a chemotherapeutic target.

Human malaria is caused by four species of the parasitic protozoan genus Plasmodium. Of these four species, Plasmodium falciparum is responsible for the vast majority of the 300 -500 million episodes of malaria worldwide and accounts for 0.7-2.7 million annual deaths (1). Given the genetic flexibility and the resulting rapid development of resistance to almost every drug, a comprehensive understanding of plasmodial metabolic pathways is essential for the development of new chemotherapeutic strategies.
Millions of years ago, an ancestor of the phylum Apicomplexa gained a plastid by secondary endosymbiosis of a photosyn-thetic eukaryote (2,3). This chloroplast was retained, and although some chloroplastid genes were lost during evolution, many were transferred to the Apicomplexan nucleus, ultimately giving rise to the apicoplast organelle, a structure essential for parasite survival (4).
In the case of malaria parasites, especially the most virulent species, P. falciparum, a series of new "plantlike" enzymes was recently discovered. Some of these enzymes are associated with the apicoplast (5), whereas the nature of the others and the pathways they are involved in remain elusive to current bioinformatics approaches.
The plant and algae plastids are the site for many essential biochemical pathways; some of them were already found in P. falciparum. It is possible that other metabolic routes were retained and incorporated in the parasite metabolism (6).
Among these biochemical pathways, the carotenoid biosynthesis is an attractive target for investigation, because it is essential in algae, higher plants, bacteria, and fungi but absent in mammals, and its products are involved in many important metabolic functions (7).
All carotenoids are derived from the isoprenoid biosynthesis pathway and possess a polyisoprenoid structure, a long conjugated chain of double bonds, and an almost bilateral symmetry around the central double bond. Their biosynthesis starts with the condensation of two molecules of geranylgeranyl pyrophosphate (GGPP) 4 to form phytoene, the initial C40 carotenoid skeleton. Different carotenoids are derived essentially by modifications in the base structure such as cyclization of the end groups and by introduction of oxygen functions, resulting in their characteristic colors and antioxidant properties (8).
Our group has previously demonstrated (9) that intraerythrocytic stages of P. falciparum biosynthesize dolichol of 11-12 isoprenic units and certain unknown compounds when [1-3 H]geranylgeranyl pyrophosphate was used as a metabolic precursor (10). Given that plant and algae plastids are sites for many essential biochemical pathways, we asked if P. falciparum was able to produce carotenoids, once these compounds have GGPP as precursor (Fig. 1).
Using metabolic labeling with radioisotopes, in vitro inhibition tests with a specific inhibitor of plant carotenoid biosynthesis, and bioinformatics analysis, we demonstrated, for the first time, that the intraerythrocytic stages of P. falciparum in fact synthesize carotenoid compounds. The first enzyme of the pathway of the carotenoid biosynthesis was cloned, expressed, and biochemically characterized. The results introduced here not only contribute to the understanding of the evolution of the P. falciparum but also introduce a possible new point of study for treatments of malaria.

Reagents
Percoll was purchased from Amersham Biosciences. [

Plasmodium Falciparum Culture
The P. falciparum clone 3D7 was cultured in vitro according to Trager and Jensen (11) with the modification that parasites were grown in a 40-ml volume in an atmosphere of 5.05% CO 2 , 4.93% O 2 , and 90.2% N 2 . In some experiments, an atmosphere of 5.05% CO 2 , 20% O 2 , and 74.95% N 2 was applied. The cultures were initially synchronized in ring stage (1-10 h after invasion) by two treatments with 5% (w/v) D-sorbitol solution in water, and the parasites were maintained in culture until the development of trophozoite (20 -24 h after reinvasion) or schizont (30 -35 h after reinvasion) stages. Parasite development and multiplication were monitored by microscopic evaluation of Giemsa-stained thin smears.

Metabolic Labeling
Synchronous cultures of P. falciparum were labeled in the ring, trophozoite, or schizont stages with [1-3 H]GGPP (3.125 Ci/ml) in normal RPMI 1640 medium for 16 h. Subsequently, parasites were then recovered, showing 25-30% parasitemia, in trophozoite, schizont, and ring stages, respectively. After labeling, the parasites cultures were centrifuged at 2,000 ϫ g for 10 min at room temperature, and each pellet was resuspended in 10 ml of phosphate-buffered saline (0.007 M Na 2 HPO 4 , 0.01 M NaH 2 PO 4 , pH 7.4, 0.15 M NaCl) containing 0.1% of saponin in order to separate the parasites from erythrocytes. After three washes with phosphate-buffered saline at 10,000 ϫ g for 10 min, the parasites were lyophilized and stored in liquid nitrogen. Another protocol described by us (12) was used to evaluate protein biosynthesis in synchronous cultures of P. falciparum.

Carotenoid Extraction
Lyophilized parasites were extracted with four volumes of ice-cold acetone three times followed by centrifugation at 8,000 ϫ g for 5 min. The pooled extracts were dried under a nitrogen stream and stored in liquid nitrogen (13).

Reverse Phase High Performance Liquid Chromatography (RP-HPLC)
Protocol I-The acetone extracts of each parasite stage obtained from metabolically labeled parasites were resuspended in 300 l of acetonitrile, filtered through a 0.45-m nylon filter (Advantec MFS, Inc., Dublin, CA), and then analyzed with a Phenomenex Luna C18 column (250 ϫ 4.6 ϫ 5 m). A gradient elution system was used, with acetonitrile/ ethyl acetate/water (88:2:10, v/v/v) as solvent A and acetonitrile/ethyl acetate/water (85:15:0) as solvent B. We applied the following gradient program: 0 -15 min, 0 -100% B; 15-45 min, 100% B; 45-50 min, 100 to 0% B; 50 -55 min, 0% B. The flow rate was adjusted to 1 ml/min, and the column was maintained at 29°C. The UV detector was set at 450 nm, and fractions were collected at 1-min intervals using a Gilson HPLC 322 pump (Gilson, Villiers-le-Bel, France) and also a gradient module connected to a 152 UV-visible detector, an 831 temperature regulator, and an FC203B fraction collector. We co-injected the samples with carotenoid extracts from six Amazon fruits previously characterized (13).
Protocol II-The acetone extracts of each parasite stage were resuspended in 20 l of methyl tertiary-butyl ether, and then analyzed using a YMC C 30 polymeric column (4.6 ϫ 250 mm, 3 m and/or 5 m) (YMC Inc.). We utilized a gradient elution system with methanol, 0.1% triethylamine (v/v) as solvent A and methyl tertiary-butyl ether as solvent B. The following gradient program was applied: 0 -30 min, 95-70% A; 30 -50 min, 70 -50% A; 50 -80 min, 50% A. The flow rate was set to 1 ml/min, and the column was maintained at 30°C. Fractions were collected at 1-min intervals. In the experiments with radioactively labeled parasites, 20 l of carotenoid extract mixture (obtained from Amazon fruits) in methyl tertiary-butyl ether that served as standard were co-injected. We carried out the analyses in an HPLC-photodiode array (PDA) equipped with a model 600 quaternary solvent delivery system (Waters, Milford, MA) and an on-line degasser, a Rheodyne injection valve with a 20-l loop, and an external oven coupled to the model 996 PDA detector (Waters). We used the Millenium Waters software for data acquisition and processing. The PDA was set at 450, 346, and 288 nm for the analysis of carotenoids (13).
For purification and identification of dephosphorylated polyisoprenoid products (40 or more carbons), the other compounds made by PfB0130w, we carried out HPLC as described previously (14,15).
In all experiments with radioactively labeled parasites, the resulting fractions were dried using a SpeedVac, resuspended in 500 l of liquid scintillation mixture (PerkinElmer Life Sciences), and monitored with a Beckman LS 5000 TB ␤-counter.

Mass Spectrometry Analysis
We used a Finnigan LCQ-Duo ion trap mass spectrometer (ThermoFinnigan, San Jose, CA) to collect the ion trap electrospray ionization-mass spectrometry (ESI-MS) and electrospray ionization-tandem mass spectrometry (ESI-MS/MS) data. The samples, which were purified by RP-HPLC (Protocol I) and corresponded to the all-trans-␤-carotene and all-trans-lutein standards, were dried under a nitrogen stream and resuspended in 30 l of chloroform/methanol (1:1, v/v), containing 2 mM lithium iodide. The samples (10 l) were directly infused into the ESI probe across a 10-l loop using an OMNIFIT N 2 pressure system (Omnifit Ltd., Cambridge, UK) with 10 p.s.i. pressure at a 10 l/min flow rate. The all-trans-␤-carotene and all-trans-lutein standards were run in the ESI positive ion mode with spray voltage, capillary voltage, and capillary temperature set to 4.52 kV, 17 V, and 250.0°C, respectively. For ESI-MS/MS, relative collision energy of 40% (2 eV) was applied, and the sheath (N 2 ) and collision helium gas pressures were set to 1.5 millitorrs and 4 p.s.i., respectively. The MS spectra were acquired in full ion as well as in single ion monitoring modes.
The lithium iodide was used to improve the ionization of the carotenoid molecules in the ESI positive mode. Different from what was observed for the noncyclic polyisoprenoid compounds (16), the ions formed from carotenoid compounds did not present an addition of 7 Da, corresponding to the molecular mass of lithium. This ionization process improved the limit of detection of all-trans-lutein and all-trans-␤-carotene from 1 nM and 2.5 nM, respectively, to ϳ100 pM.

HPLC and LC-Atmospheric Pressure Chemical Ionization (APCI)-MS/MS Analyses
Carotenoids were analyzed using C 30 columns, and the carotenoid standards used were obtained from six Amazon fruits previously characterized by LC-PDA and LC-PDA-APCI-MS/MS comparing with HPLC grade standards from DSM Nutritional Products (Basel, Switzerland) to guarantee the identity of the compounds present in each chromatographic fraction (13). Analyses of all-trans-␤-carotene and alltrans-lutein from unlabeled schizont stages were performed in the same way as described above.

Inhibition Tests with the Herbicide Norflurazon and Chloroquine
We first diluted norflurazon in ethanol and then in RPMI medium to the following concentrations: 1000, 100, 10, 1, and 0.1 M. Chloroquine was diluted directly in the RPMI 1640 medium, yielding concentrations of 20, 10, 5, 1, and 0.5 nM. Inhibition tests were carried out in flat-bottomed microtiter plates. We applied the method proposed by Desjardins et al. (17) to determine the 50% inhibitory concentrations (IC 50 values). The IC 50 chloroquine value used was lower than as the one described by Cremer et al. (18).
The IC 50 values for growth inhibition were calculated by Probit Analysis (Minitab Statistical Software 13.30 TM; Minitab Inc.). After determination of the IC 50 values for norflurazon (25 Ϯ 1.0 M) and chloroquine (10 Ϯ 0.5 nM), the parasites in ring stage were treated with either 25 M norflurazon or 10 nM chloroquine for 24 h and were metabolically labeled with [1-3 H]GGPP in the presence of the drug. Carotenoid compounds were then extracted and analyzed by HPLC (Protocol II) to determine the pattern of carotenoid compounds synthesized by the treated and untreated parasites.

Rescue Assay
Lycopene was solubilized in THF, 10 mM BHT and then diluted to 0.2 mM with human serum. Next, we added the THF/ human serum-lycopene preparation to synchronous P. falciparum cultures treated with 25 M norflurazon to yield a final lycopene concentration of 200 nM; the parasitemia was read every 24 h.

Enzyme Assays of Crude Parasite Extracts
We diluted 1-2 mg/ml Triton X-100 schizont extracts in an incubation buffer containing 2 mM NADP, 0.02 mM FAD, and 10 M [1-3 H]GGPP in a total volume of 1 ml. The extracts were then incubated at 30°C for 3 h in the dark with agitation (19).

Cloning, Expression, and Purification of PfB0130w Product Gene
P. falciparum genomic DNA was PCR-amplified using primers designed according to the complete PfB0130w sequence (GenBank TM accession number GI3845103-AAC71816), with the introduction of BamHI restriction sites (sense, CCG GAT CC ATG GTT CAC CTA AGT AAA AG; antisense, CCG GAT CCT CAT TTG AGG TTT CTT GAT AAC). The 1.6-kb amplicon was initially cloned into pGEMT-easy vector (Promega) and subcloned into the BamHI site of pRSETB (Invitrogen). The resulting plasmid was then transformed in Escherichia coli BL21-CodonPlus TM (DE3)-RIL (Stratagene). Protein expression was induced at A 600 of 1.0 with 0.3 mM isopropyl 1-thio-␤-D-galactopyranoside overnight at 25°C, after which bacterial cells were harvested and resuspended in lysis buffer (7 mM Na 2 HPO 4 , 1 mM KH 2 PO 4 , pH 7.2, 137 mM NaCl, 3 mM KCl, 1% Triton X-100 (v/v), 0.05 mg/ml lysozyme, and 0.2 mM phenylmethylsulfonyl fluoride). Lysis was completed with three freeze/thaw cycles, and genomic bacterial DNA was sheared by 10 passages through a G21 needle. We then purified recombinant proteins using Ni 2ϩ chelating affinity chromatography. Twelve different concentrations (1-300 M) of substrate [1-3 H]GGPP were examined to determine kinetic parameters of recombinant PfB0130w. A time kinetic study was also performed at 15 min, 30 min, 1 h, 2 h, 3 h, and 15 h. K m and V max values (means and S.E.) were obtained by the least square method using Enzfitter software (Elsevier-Biosoft, Cambridge, UK). Radiometric assays were carried out by previously cited modified methods (21).

Bioinformatics Analyses
To identify P. falciparum candidate carotenoid genes, we compared carotenoid biosynthesis-related genes (retrieved from GenBank TM ) against a local data base of Plasmodiumspecific genes (Plasmodium berghei, Plasmodium chabaudi, Plasmodium falciparum, Plasmodium vivax, and Plasmodium yoelii sequences; PlasmoDB, version 5.0) using BLAST and hidden Markov model (HMM) approaches. BLAST cut-off was Ͻ E value of 10 Ϫ5 . HMMs of known protein domains implicated in carotenoid biosynthesis were downloaded from Pfam (version 20.0; available on the World Wide Web). We used HMMs to search putative sequences against the Plasmodium local data base utilizing HMMsearch, part of the HMMER2 package. Protein sequences were aligned with the ClustalX version 1.83 program. We employed the following bioinformatic tools to predict the PfB0130w cellular location: PlasmoDB, WoLF PSORT, iPSORT, MITOPROT, Protein Prowler 1.2, and Predotar.

RESULTS
Biosynthesis of Carotenoids by the Intraerythrocytic Stages of P. falciparum-In a first approach, we tried to identify de novo synthesized carotenoids in vitro cultivated blood stage parasites. For this, 2.7 ϫ 10 9 parasites from each stage (ring, trophozoite, and schizont) were metabolically labeled with [1-3 H]GGPP, lysed by saponin treatment and extracted with acetone, and then analyzed on a C 18 reverse phase column (Protocol I). A radioactive fraction with retention times coincident with alltrans-lutein standard (12 min) was detected in all intraerythrocytic stages, whereas the radioactive fraction coincident with all-trans-␤-carotene standard (38 min) was only detected in schizont stage parasites ( Fig. 2A). The commercial standards were co-chromatographed with the parasite samples and identified by their respective UV-visible spectra (450 nm for both standards). To verify the molecular identity of the compounds present in these two fractions, the products of the HPLC analysis of the acetone extracts of 3.0 ϫ 10 10 unlabeled schizont parasites were analyzed by ESI-MS and ESI-MS/MS (Fig. 2B). We chose to analyze only the schizont stage by mass spectrometry, because this parasite form exhibited the highest levels of carotenoid biosynthesis. In these analyses, identical ionization profiles were observed in the parasite samples and in the standards, with all-trans-lutein identified by the [M Ϫ H 2 O] ϩ ion at mass-to-charge ratio (m/z) 550 and all-trans-␤-carotene by the [M] ϩ ion at m/z 536 (Fig. 2B). The molecular structures were confirmed by obtaining matching dissociation patterns when comparing the ESI-MS/MS spectra of the ions with m/z 550 and m/z 536 from P. falciparum with those of the all-translutein and all-trans-␤-carotene standards, respectively (Fig.  2B). Importantly, these compounds were de novo synthesized by the parasites, since they were not detected by mass spectrometry either in RPMI or in normal red blood cells used in the parasite culture (Fig. S1).
Once the biosynthesis of all-trans-lutein and all-trans-␤-carotene was observed in the intraerythrocytic stages of P. falciparum, we investigated the possible presence of other carotenoid compounds in these parasite forms. Using another HPLC method (Protocol II) with a C 30 reverse phase column, putatively ideal for the analysis of complex mixtures of carotenoids (22), acetone extracts of the three intraerythrocytic stages of P. falciparum (2.7 ϫ 10 9 parasites of each stage), metabolically labeled in vitro with [1-3 H]GGPP and subsequently isolated from erythrocytes by treatment with 0.1% of saponin, were analyzed. Radioactive fractions with retention times coincident with all-trans-lutein (15 min), phytoene (18 min), all-transphytofluene (22 min), all-trans-␤-carotene (35 min), cis-␥-carotene (39 min), cis-␤-zeacarotene (42 min), and cis-␦-carotene (44 min) standards (Fig. 3A) were observed. The highest quantities of these compounds were once again detected in the schizont stage parasites. As before, we co-chromatographed the mixture of carotenoid standards used with the parasite samples and then identified the compounds by their respective UV-visible spectra (450 nm for all-trans-lutein, all-trans-␤-carotene, cis-␥-carotene, cis-␤-zeacarotene, and cis-␦-carotene; 350 nm for all-trans-phytofluene and 286 nm for phytoene). This carotenoid standard mixture was obtained from six native Amazonian fruits, and the compounds present in each chromatographic fraction of this mixture were previously characterized by LC-PDA and LC-PDA-APCI-MS/MS (13). Radioactive fractions with retention times coincident with neurosporene (51 min) and all-trans-lycopene (77 min) standards were additionally detected in schizont stages (Fig. S2). No radioactive fractions coinciding with the carotenoid standards used were found in the acetone extracts from uninfected erythrocytes labeled with [1-3 H]GGPP. Under the above chromatographic conditions, polyisoprenoids have different retention times from those of carotenoids as described in Table S1.
Since all-trans-lutein and all-trans-␤-carotene compounds had already been characterized using HPLC-C 18 and ESI-MS/MS analyses, we then attempted to confirm the identity of these two compounds by a new mass spectrometry approach after the HPLC-C 30 analysis. For this, 5 ϫ 10 10 unlabeled schizont parasites, isolated from erythrocytes by treatment with 0.1% saponin, were subjected to carotenoid extraction with ace-  (Fig. 3B). The molecular structures were confirmed by obtaining matching dissociation patterns of the APCI-MS/MS spectra of the ions with m/z 569 and m/z 537 found in P. falciparum and those in the all-trans-lutein and all-trans-␤-carotene standards, respectively (Fig. 3B). Again, none of these compounds were detected in uninfected erythrocytes and RPMI medium (Fig. S3).
Enzymatic Evidence for the Carotenoid Pathway in P. falciparum-In order to confirm the presence of carotenoid biosynthesis in the intraerythrocytic stages of P. falciparum, we performed an in vitro enzymatic activity assay using Triton X-100 crude extracts of 5 ϫ 10 10 schizont parasites separated from erythrocytes by treatment with 0.1% saponin. Following the addition of [1-3 H]GGPP as a substrate and subsequent HPLC analysis (C 30 column, 5 m; Protocol II) of the acetone extracts, radioactive fractions with retention times coincident to all-trans-lutein (14 min), phytoene (20 min), all-trans-phytofluene (24 min), all-trans-␤-carotene (35 min), cis-␥-carotene (39 min), and cis-␦-carotene (44 min) standards were observed (Fig. 3C). Under identical experimental conditions, extracts of uninfected erythrocytes did not lead to the production of any radioactive fractions (data not shown).
Norflurazon Treatment Inhibits the P. falciparum Carotenoid Biosynthesis-Next, we investigated whether carotenoid biosynthesis is essential for parasite survival or solely represents an evolutionary remnant. To this end, parasites were treated with norflurazon, a well known bleaching herbicide that inhibits carotenoid biosynthesis in higher plants and microalgae. Parasite growth was inhibited in a concentration-dependent manner ( Fig. 4A) with an IC 50 value of 25 Ϯ 1.0 M at 48 h of treatment. At this IC 50 , no inhibition of protein biosynthesis was observed (Fig. S4). Interestingly, the inhibitory effect of norflurazon was partially reversed (77 Ϯ 8.0%) by the addition of 200 nM lycopene (Fig. 4B), a downstream product in the carotenoid pathway. This finding suggests that the carotenoid biosynthesis pathway is commonly used by the parasite and that norflurazon affects it specifically. The effect of norflurazon on carotenoid biosynthesis in the schizont stage was further investigated. Consistent with the hypothesis that blockade of the phytoene desaturase enzyme leads to accumulation of precursors and depletion of downstream products, the treatment of parasites with 25 M norflurazon increased phytoene content and diminished other carotenoid compound levels (Fig. 4C).
Characterization of the P. falciparum Phytoene Synthase, a Bifunctional Enzyme-Since no carotenoid synthesis-related genes were specified in the P. falciparum genome, we attempted to identify possible Plasmodium genes associated with the carotenoid biosynthesis pathway. An exhaustive bioinformatic search, based on HMMs and BLAST analyses, identi-fied PfB0130w as a potential candidate for the P. falciparum phytoene synthase. Interestingly, our group previously characterized PfB0130w as an octaprenyl pyrophosphate synthase  (14). However, no candidates for other carotenoid synthesisassociated enzymes could be unambiguously identified (data not shown). Similar results were recently obtained by Nagamune et al. (23) in an attempt to identify abscisic acid synthesis genes in T. gondii. The protein sequence that most corresponded to PfB0130w was the predicted phytoene synthase of the purple bacterium Rubrivivax gelatinosus (NCBI accession number BAA94032, 21% amino acid identity after pairwise alignment) (Fig. S5). Although not confirmed by biochemical evidence, the Rubrivivax protein was predicted to contain a trans-isoprenyl diphosphate synthase head-to-tail domain similar to the one found in PfB0130w protein but distinct from the head-to-head domains present in other phytoene synthases (24).
To verify the existence of a phytoene synthase activity of the PfB0130w protein, we conducted in vitro enzymatic assays using [1-3 H]GGPP as a substrate, followed by HPLC analysis (Protocol II). Only one radioactive fraction with a retention time coincident with the phytoene standard was observed (Fig.  5A). LC-APCI-MS/MS analysis of the product resulting from the enzymatic reaction using GGPP revealed the presence of the phytoene [M ϩ H] ϩ ion at m/z 545 (Fig. 5B). We detected the same [M ϩ H] ϩ ion when subjecting the phytoene standard to LC-APCI-MS (Fig. 5B). The molecular identity was confirmed by comparing the LC-APCI-MS/MS of the ion at m/z 545 from the enzymatic reaction (Fig. 5B) with that of the standard. Both spectra yielded identical and structurally diagnostic dissociation profiles.
When the truncated version of PfB0130w recombinant protein, previously described by our group (14), was used with the same substrate [1-3 H]GGPP and under identical reaction conditions as for the analysis of phytoene synthase activity, no products with retention times corresponding to phytoene were found, indicating that the carboxyl-terminal domain of the enzyme is essential for the phytoene synthase function (data not shown). To investigate if the octaprenyl pyrophosphate synthase activity was present in the full-length PfB0130w gene product, we incubated the enzyme with farnesyl pyrophosphate ammonium salt and [ 14 C]IPP under the conditions optimal for polyprenol synthesis and then detected products containing 40, 45, and 50 carbons (Fig. S6), a finding similar to that obtained with the truncated enzyme. The kinetic experiments showed that upon increase of the GGPP concentration in the full-length PfB0130w gene product assay mixture, a corresponding increase of the phytoene formation was observed, and saturation was achieved at GGPP concentrations above 300 M. The Lineweaver-Burk plot of substrate-velocity data yielded a linear relationship with an apparent K m of 21.7 Ϯ 3.3 M for GGPP and an apparent V max of 34.6 Ϯ 5.2 nmol mg Ϫ1 protein h Ϫ1 (Fig.  6). Since the recombinant full-length enzyme PfB0130w shows specific activity and Michaelis-Menten kinetics similar to those of the other phytoene synthases (20,21,25), a time kinetic study was performed. Similar to results published by Iwata-Reuyl et al. (20), phytoene production was detected after 15 min of incubation, and no significant increase after this was observed.
If the phytoene synthase activity of the PfB0130w protein is analogous to the one found in the carotenoid pathway of plants, then we expect this protein to be located in the apicoplast of the parasite. However, when scanning the peptide sequence with five different programs for intracellular location prediction, the related protein is localized in a mitochondrial compartment. In order to identify the localization, we conducted immunofluorescence assays using antibodies against the truncated version of this protein (14) (Fig. S7). The PfB0130w protein was localized to the cytoplasm of P. falciparum late trophozoite stages, and the fluorescence partially overlapped with that occupied by the mitochondrial marker and anti-GFP (apicoplast) (Fig. S8).
Probable Antioxidant Action of Carotenoids in P. falciparum-To test if the carotenoids provide essential antioxidant protection for P. falciparum parasites, we treated the parasites with chloroquine (10 nM for 24 h), a drug that inhibits ferriprotoporphyrin IX degradation, causing the generation of reactive oxy- gen species (26). We speculated that the rise in reactive oxygen species levels would enhance the biosynthesis of carotenoids. As expected, chloroquine treatment in fact increased carotenoid biosynthesis (Fig. 7A), suggesting that carotenoids are probably involved in the antioxidative defense system of P. falciparum.
Additional experiments were conducted to reinforce this hypothesis. For this, cultures of P. falciparum were treated for 48 h with 25 M of norflurazon (IC 50 ) in a 20% O 2 atmosphere. In these conditions, the parasitemia decreased around 65% when compared with parasites treated in the same conditions but maintained in a 5% O 2 atmosphere (Fig.  7B). As a control, cultures of P. falciparum were treated with the unrelated drug at its IC 50 of 25 M risedronate 2-(3pyridyl)-1-hydroxyethane-1,1-biphosphonate (27), and no effect was observed in both culture conditions (data not shown).

DISCUSSION
In this study, we show for the first time that the biosynthesis of carotenoids is functionally active in the intraerythrocytic stages of P. falciparum, representing another plantlike pathway present in these parasites.
The unambiguous identification of molecules is commonly conducted by the usage of at least two biochemical analysis procedures. The carotenoids biosynthesized by the intraerythrocytic stages of P. falciparum were shown by metabolic labeling with the direct precursor [1-3 H]GGPP and identified by two HPLC methods, suitable for these type of molecules (28), and confirmed by ESI-MS/MS and LC-APCI-MS/MS analyses, excluding any uncertainty about the molecular nature of the detected compounds. Apparently, the schizont stages contained the highest quantities of carotenoids when compared with rings and trophozoites. This indicates that carotenoid synthesis starts in ring stage and accumulates in schizont stage. Importantly, neither of these compounds was detected in uninfected erythrocytes or in RPMI culture medium and subsequent tests in P. falciparum extracts showed that the parasite comes with the machinery to synthesize carotenoids.
Our results carry us to an important question. Are the carotenoids important for the parasite metabolism, or do they only represent an evolutionary vestige? To verify this, we treated the parasites with norflurazon, a well known bleaching herbicide that inhibits carotenoids biosynthesis in higher plants and microalgae by competitive inhibition of the phytoene desaturase (29). Benz-Amotz et al. (30) colleagues showed that norflurazon treatment of the halotolerant green alga Dunaliella bardawil blocks the production of all-trans-␤-carotene and provokes, in certain conditions, the accumulation of massive amounts of phytoene, a result that was also observed in other carotenogenic organisms.
In our hands, P. falciparum reacted quite comparably with D. baradawil and showed growth inhibition upon norflurazon treatment. Likewise, the inhibition could be partially reverted by the addition of lycopene, which seems to be readily taken up, thus providing the products of the norflurazon-inhibited step in the carotenoid pathway. The norflurazon inhibition was not completely reversed, most likely because lycopene is unstable in solution, and the solvent (THF) indicated for carotenoid delivery to cells is cytotoxic (31).
The genes related to several important activities exerted by Plasmodium, such as transcriptional control (32) or the shikimate metabolism, continue to be elusive to bioinformatic approaches (33). We tried to identify candidate sequences/ genes for the carotenoid biosynthesis in P. falciparum, search- ing for known sequences involved in this pathway in other organisms against a Plasmodium local data base using BLAST (34) and HMMER (35) methods. Our in silico analyses suggest the presence of a candidate for phytoene synthesis. Intriguingly, the candidate gene encoding the enzyme phytoene synthase that synthesizes the first product of the carotenoid biosynthesis, phytoene, was previous characterized by our group as an octaprenyl pyrophosphate synthase (14). The plasmodial enzyme is a rare example of a carotenogenic enzyme with a continuous line of evolution from archaea to bacteria (via cyanobacteria) and plants (36,37) and contains two activities. Although the NH 2 -terminal 344 amino acids are sufficient for the octaprenyl activity (14), the whole polypeptide exerts double activity as octaprenyl pyrophosphate synthase and phytoene synthase. Taking into account that the phytoene synthase of Rubrivivax gelatinosus was only predicted to be a trans-isoprenyl diphosphate synthase of the head-to-tail type, our results provide the first biochemical evidence that a protein with this configuration is able to synthesize phytoene.
The recombinant full-length enzyme pPfB0130w has a similar specific activity when compared with other phytoene synthases. The K m value of the Erwinia uredorova phytoene synthase was 41 M (21), whereas the enzyme of Capsicum annum had a K m of 0.3 M (38). It appears that the K m value for this kind of enzyme is very variable; the value found for the P. falciparum enzyme lies between those described in the literature.
Our data indicate that this enzyme is actually a bifunctional enzyme, exhibiting octaprenyl pyrophosphate synthase and phytoene synthase activities, analogous to the other bifunctional enzymes of the isoprenoid pathway found in Apicomplexan parasites (39).
The PfB0130w protein was predicted to have a mitochondrial localization; however, the possibility of co-localization with the apicoplast was unexpected, because computational analysis does not predict a plastid targeting sequence at the NH 2 termini of the proteins. Although the resolution of immu-nofluorescence assays did not permit a conclusive result, we suggest that the protein may localize to both or to the interface of these organelles. In P. falciparum, the mitochondrion is closely associated with the apicoplast, and enzymes of some biochemical pathways localize to both organelles (40,41). Also, mitochondrial metabolism is not common in this parasite; once this organelle does not present crests, it is not associated with the energy production, and the function of the tricarboxylic acid cycle in the mitochondrion is unclear (40). This allows for a possibility that carotenoid biosynthesis may take place in both   organelles, like the heme biosynthesis that occurs in both the mitochondrion and plastids (37).
If one considers that the first enzymatic function of the PfB0130w protein results in an isoprenic compound of 40 carbons, probably the isoprenic chain of P. falciparum ubiquinone, the localization of this enzyme associated with the mitochondrial membrane would be expected. A problem arises regarding the phytoene synthase activity, which in plants is normally localized to chloroplasts (42). It may be questioned if this reaction can occur in the mitochondrial environment. The daffodil model of carotenoid biosynthesis can answer this question (43), and according to this model, the biosynthesis of carotenoids takes place in a membrane-associated enzyme complex, but different reactions and enzymes require different microenvironmental conditions, including membrane-resident electron acceptors and donators and anaerobic environments (44). We speculate that the daffodil model could represent a good approach to describe the carotenoid biosynthesis in P. falciparum; nevertheless, the detection of other carotenoid-related enzymes and knowledge about their exact localization are indispensable prerequisites for the description of a model.
Regarding the timing of expression, the PfB0130w protein is present in trophozoites and schizonts, coinciding with previous findings, which showed that (i) PfB0130w transcription occurred mainly in ring and trophozoite stages (14) and (ii) the highest levels of carotenoid (this study) and other isoprenoid derivates (12,45) were found in schizont stages.
The carotenoids apparently are important for the parasite metabolism, but what are their functions? Considering that the main purpose of carotenoid compounds, even in the photosynthesis, is antioxidant, it has been suggested that they possess specific tasks in the antioxidant network, such as protecting lipophilic compartments or scavenging reactive species generated in photooxidative processes, neutralizing mainly singlet molecular oxygen and peroxyl radicals (8). In the case of P. falciparum, during the evolutionary progression from mixotrophy to parasitism, the ancestral pathogen probably encountered an increase in oxidative stress. Perhaps, as an adaptive response to this increase in oxidative stress, Plasmodium gradually developed into a fermentative organism (46). On the other hand, the parasites live in a pro-oxidant environment that contains oxygen and iron, the key prerequisite for the formation of reactive oxygen species via the Fenton reaction, and is not surprising that P. falciparum are heavily dependent on efficient antioxidant systems (47)(48)(49)(50)(51)(52).
To verify if carotenoids could be involved in the antioxidant systems of the parasite, we induced oxidative stress by chloroquine treatment, since Ginsburg and Krugliak (53) showed that chloroquine inhibits the ferroprotoporphyrin IX polymerization, causing the generation of reactive oxygen species due to the Fe 2ϩ ion, which is not detoxified.
The treatment with chloroquine demonstrated an increase in the carotenoid levels, indicating that carotenoids could be involved in the antioxidative defenses of P. falciparum. Additionally, a more efficient inhibitory effect of norflurazon on the growth of parasites under higher levels of O 2 (20%) and no differences with residronate treatment support this hypothesis. Obviously, these results are not sufficient to prove the involve-ment of carotenoids in the control of oxidative stress in this parasite but indicate a possible function. The possible presence of the phytoene synthase associated with the mitochondrion and the other carotenoid-related enzymes in the apicoplast points to an antioxidant function of the carotenoids, since the mitochondrion and the apicoplast are the key organelles in antioxidant defenses for P. falciparum (54,55).
Another possible function for carotenoid compounds is the coordination of plastid and nuclear gene expression. Carotenoids have been previously suggested to function as intracellular messengers, coordinating plastid and nuclear gene expression (56) and/or membrane structure modification (57). It was shown that carotenoids like all-trans-lutein and all-trans-␤carotene can be incorporated in membranes, modifying their physical properties (57). These two functions could be hypothesized in the parasite metabolism, since practically all proteins that have their function in the apicoplast are nuclear encoded, and the constitution of the parasite membranes is not clear, because P. falciparum do not synthesize cholesterol (58).
Despite all of these important hypotheses for carotenoid function in P. falciparum metabolism, it is still possible that carotenoid biosynthesis in this parasite may only serve to produce carotenoid hormones like abscisic acid, as was demonstrated in T. gondii (23), or other downstream compounds like retinoids, potentially leaving no specific function for the carotenoid intermediates. We explored the possibility of retinoid synthesis in the intraerythrocytic stages of P. falciparum but found the parasites to be unable to produce these compounds (data not shown).
In conclusion, we have shown that intraerythrocytic stages of P. falciparum have an active carotenoid biosynthesis pathway and that it plays an important role in the parasite's metabolism. Given that carotenoid biosynthesis is absent in humans but present in P. falciparum and, moreover, contains an unusual protein phytoene synthase characterized in this work, we speculate that this pathway could be exploited as a new target for antimalarial drugs. Sequence data from additional organisms, functional studies, improved bioinformatic screening approaches, and additional biochemical evidence may further reveal the potential of this new pathway in this important human parasite.