Type II fatty acid synthesis is essential only for malaria parasite late liver stage development

1Malaria parasites are protists belonging to the genus Plasmodium. They are obligate intracellular parasites that have two distinct replicating life cycle forms in the mammalian host. A massive one-time replication occurs in the liver after inoculation of sporozoite stages by the bite of an infected mosquito and results in the production and release of tens of thousands infectious exo-erythrocytic merozoites. These merozoites infect red blood cells and initiate the cyclic replication that occurs within the blood stream. Blood stage infection leads to malaria disease with Plasmodium falciparum alone afflicting more than 500 million people annually. Plasmodium replication in red blood cells produces between 8 and 36 merozoites with each invasive cycle whereas one-time replication in the infected hepatocyte produces up to 40 000 merozoites – an ∼2000-fold difference. It is currently not well understood to what extent malaria parasites rely on parasitic scavenging of nutrients versus intrinsic synthesis for growth and replication.

2Lipids are not only essential but are one of the most abundant components of all organisms and the malaria parasite needs a plentiful supply of lipids – specifically fatty acids for the membrane biogenesis necessary for invasive stage formation. Plasmodium parasites were initially assumed to lack the ability to synthesize their own fatty acids and thus rely on their hosts for lipid scavenging. However, this model came into question with the discovery of the apicoplast, a relict plastid organelle of Plasmodium. Plant and algal plastids harbour several key biosynthetic pathways and the sequencing of the P. falciparum genome coupled with a detailed analysis of the proteins of known function that were targeted to the apicoplast allowed the construction of an apicoplast-specific metabolic map. The apicoplast is of cyanobacterial origin and one such apicoplast-targeted pathway is bacterial-like type II fatty acid synthesis (FAS II), a de novo pathway by which Plasmodium can synthesize fatty acids from derivatives of acetate and malonate. The fatty acid chain extension step of FAS II is catalysed by four key enzymes – FabB/F, FabG, FabI and FabZ and the substrate/product of each reaction is covalently bound to the acyl carrier protein (ACP) cofactor (Fig. 1A). Conversely, the mammalian FAS I pathway utilizes a single enzyme complex and is not present in Plasmodium based on genome sequence analysis. This is not the case for all apicomplexan parasites – the genome of Toxoplasma gondii encodes both FAS I and FAS II enzymes, Cryptosporidium parvum has FAS I enzyme whereas Theileria annulata does not harbour either FAS I or FAS II pathways. Deletion of ACP from T. gondii has demonstrated that apicoplast fatty acid synthesis is essential for organelle biogenesis and parasite survival in this parasite.

3The four Plasmodium FAS II enzymes are promising drug targets because they are of bacterial origin. The P. falciparum enzymes have been expressed in vitro and used to reconstitute the elongation module of FAS II. The in vitro system mimicked the in vivo machinery and known inhibitors of the enzymes of the elongation module caused the expected accumulation of intermediates. Thus, Plasmodium possesses a functional FAS II pathway. An early study identified a Plasmodium FabI and showed that the FabI inhibitor triclosan kills blood stage parasites and subsequently a significant effort has been undertaken to develop blood stage FAS II inhibitors to treat malaria. Although the data suggested that FAS II is necessary for intra-erythrocytic replication, the expression of FAS II enzymes has not been studied throughout the complex infection cycle of the parasite and their importance in parasite progression throughout the life cycle remains unknown.

4We recently carried out a liver stage transcriptome and proteome analysis in the model rodent malaria parasite Plasmodium yoelii and observed that (i) the transcription of FAS II genes was increased in liver stages when compared with blood stages; (ii) FAS II enzymes were present in the liver stage proteome; and (iii) hexachlorophene, an inhibitor of FabG, was able to inhibit liver stage development in vitro. These results suggested that FAS II might be important for parasite liver infection. Here, we show that expression of FAS II only occurs during the pre-erythrocytic phases of parasite infection and this allowed unprecedented imaging of the sporozoite and liver stage apicoplast. Strikingly, gene knockouts of FabB/F and FabZ, two of the enzymes involved in fatty acid synthesis demonstrated that FAS II is critical for normal liver stage development but not for blood stage or mosquito stage development. Knockout parasites did not form the first generation of invasive merozoites, which are the critical parasite transition stages to move from the liver into the blood stream and initiate red blood cell infection. In addition, gene knockout of FabI from P. falciparum, a further enzyme involved in FAS II, demonstrated that FAS II is not critical for P. falciparum blood stage replication.

5We used quantitative RT-PCR (qPCR) to show upregulation of FAS II genes in liver stages because our previous microarray analysis had indicated preferential expression in liver stages of P. yoelii. Transcript abundance was analysed in the different P. yoelii life cycle stages for the four FAS II genes encoding the enzymes involved in fatty acid extension (Fig. 1A). For all four genes, the level of transcription was highly increased in pre-erythrocytic stages but the most consistent and substantial expression was seen in late liver stages when compared with blood stages (Fig. 1B to E). The qPCR data suggest that FAS II is induced in liver stages and might thus play an important role for liver stage development.

6To further investigate the expression of FAS II during the parasite life cycle we generated a transgenic P. yoelii line expressing a myc epitope-tagged FabI under the control of the endogenous FabI promoter (PyFabI-myc) (Fig. S1). The quadruple myc tag was fused to the carboxyl (C)-terminus of FabI and was followed by the 3′ UTR from Plasmodium berghei dihydrofolate reductase/thymidylate synthase (DHFR/TS). This expression cassette design should not alter the stage-specific expression of FabI as 3′UTRs are involved in mRNA stability, but not regulation of temporal expression. A similar strategy has previously been used to study the blood stage expression of P. falciparum subtilase 1. PyFabI-myc allowed for the visualization of FabI expression throughout the parasite life cycle using indirect immunofluorescence assays (IFA) with anti-myc antibodies. FabI-myc expression was not detectable in the developing mosquito midgut oocysts and also not in oocyst sporozoites (Fig. S2). FabI-myc expression was first detected in salivary gland sporozoites and localized to a spherical structure close to the nucleus (Fig. 2A). All P. yoelii FAS II enzymes including P. yoelii FabI possess a bipartite leader sequence that predicts import into the apicoplast (data not shown). As FAS II enzymes localize to the apicoplast in the apicomplexan T. gondii, and FabI from the related apicocomplexan Eimeria tenella was localized to the apicoplast, we assume that the FabI-myc expression we detected in salivary gland sporozoites is apicoplast specific (Fig. 2A). This is the first published data showing the Plasmodium sporozoite apicoplast. To analyse FabI expression during liver stage development, HepG2:CD81 hepatoma cells were infected with sporozoites from the PyFabI-myc line. At 7 h post infection (pi) when intracellular sporozoites initiate transformation to trophozoites, apicoplast morphology as determined by FabI-myc staining (Fig. 2B), was similar to that of salivary gland sporozoites (Fig. 2A). At 14 h pi, parasite nuclear division commenced and the apicoplast initiated its division as indicated by the dumbbell shape (Fig. 2C). By 24 h pi, the apicoplast had formed a branched lariat-shaped structure (Fig. 2D), which became more elaborate with advanced liver stage development at 30 h pi (Fig. 2E). By 40 h pi (Fig. 2F) the apicoplast had differentiated into hundreds of intertwining structures that appeared to be segregating. These results show that FabI expression is initiated in salivary gland sporozoites and that there is robust apicoplast-specific FabI expression throughout liver stage development. Interestingly, the replication of the liver stage apicoplast bears striking similarities to that seen in the apicoplast of developing P. falciparum blood stages with reference to the changing appearance of the organelle during schizogony but on a much more expansive scale. Strikingly, by 48 h pi (Fig. 2G), a time point when the P. yoelii liver stage schizont undergoes merozoite formation, FabI-myc expression was greatly reduced when compared with 40 h pi (Fig. 2F). This strongly suggests that FabI expression is downregulated shortly before or during exo-erythrocytic merozoite formation. Finally, we investigated whether FabI is expressed during asexual blood stage replication and found that FabI-myc expression was not detected in blood stages (Fig. 3). To further analyse FAS II expression, additional parasite lines were created that expressed epitope-tagged FabZ and FabG (PyFabZ-myc or PyFabG-myc). Similar expression patterns were seen throughout the life cycle when compared with FabI-myc (data not shown), exemplified by the labelling of a branched lariat-shaped structure at 24 h pi for both PyFabZ-myc and PyFabG-myc (Fig. 2H and I). Taken as a whole, the expression data led us to hypothesize that FAS II is not essential for blood stage development but might be critical for liver stage development.

7To test the hypothesis that FAS II is not essential for blood stage development, we deleted P. yoelii FabB/F, the enzyme that catalyses the condensation of malonyl-ACP with the lengthening fatty acyl-ACP formed by FabI (Fig. 1A) from the parasite genome using a double cross-over recombination strategy. We decided to delete FabB/F because it catalyses an essential step in fatty acid synthesis. PCR genotyping using specific primers pairs confirmed the recombination event and the deletion of the FabB/F gene (Fig. S3) in two cloned knockout parasite lines (fabb/f−) from two independent transfections. To assay the effect of FabB/F deletion on blood stage development, mice were intravenously (iv) injected with 1 × 103 or 1 × 106fabb/f− clone 1 parasites and blood stage development was followed over time in comparison with wild-type (WT) parasites (Fig. 4A and B). There was no significant difference in the growth rate of fabb/f− clone 1 parasites and WT parasites. Normal clearance of the parasite that is observed for this non-lethal P. yoelii strain occurred for both WT and knockout approximately 20 days pi (Fig. 4A and B). The results demonstrate that the loss of FabB/F and therefore the loss of FAS II had no deleterious effect on P. yoelii asexual blood stage replication in vivo. Furthermore, the blood stage fabb/f− parasites showed normal gametocyte development and male gamete exflagellation (data not shown). To confirm our observations with fabb/f− parasites we used similar methodologies to generate a parasite with a deletion in a second synthesis enzyme, FabZ. FabZ catalyses the dehydration of β-hydroxyacyl-ACP to form trans-2-enoyl-ACP (Fig. 1A). PCR genotyping confirmed the creation of fabz− parasites in two cloned knockout lines from independent transfections (Fig. S3). As for fabb/f− parasites, the deletion of FabZ had no deleterious effect on the growth of fabz− blood stage parasites (Fig. 4A and B). A similar blood stage phenotype from the deletion of a second enzyme in FAS II adds strength to our hypothesis that FAS II is not necessary for blood stage replication.

8After transmission of either fabb/f− or fabz− to Anopheles stephensi mosquitoes, we observed normal development of mosquito midgut oocysts, formation of oocyst sporozoites and invasion of sporozoites into the salivary glands as indicated by the enumeration of salivary gland sporozoites in comparison with WT (Table S1: data for one clone for each deletion are shown). Therefore, FAS II is not necessary for parasite development in the mosquito. Next, salivary gland sporozoites were injected iv into BALB/c mice. Mice were injected with 10 000 and 50 000 (n = 8) knockout sporozoites or with 10 000 WT sporozoites. Mice were assayed for blood stage parasitaemia every other day from day 3 post injection (pi) until day 15 by Giemsa-stained blood smears (Table 1). After 3 days, all mice injected with WT sporozoites exhibited patent blood stage parasitaemia. Strikingly, none of the mice injected with 10 000 and 50 000 fabb/f− sporozoites or fabz− sporozoites became blood stage patent and this was true for both clones of each of the knockouts. These results demonstrate that the lack of FAS II renders the pre-erythrocytic parasite unable to successfully infect the mammalian host.

9To further investigate the phenotype of the knockout parasites we chose to follow the progression of fabb/f− clone 1 parasites in the liver. BALB/c mice were iv injected with 1 × 106 sporozoites (WT or fabb/f−) and sacrificed at different time points of liver stage development. The infected livers were perfused, removed, sectioned and parasite load and development was assayed by immunofluorescence microscopy. At 12 and 24 h pi fabb/f− liver stages showed normal development that was indistinguishable from WT, i.e. they invaded hepatocytes, formed a parasitophorous vacuole (PV), transformed into trophozoites and initiated schizogony (Fig. 5A and B). However, by 44 h pi the size of the fabb/f− liver stages was significantly less than that of WT (Figs 4C and 6A). Quantification of cell size showed that more than 60% of WT liver stages at this time point had a maximum area at their widest diameter of > 1500 μm2, whereas more than 90% of the fabb/f− liver stages were less than 1500 μm2 (Fig. 4C). In addition, abnormal progression of nuclear division became apparent. Furthermore, unlike WT, fabb/f− liver stages did not show merozoite surface protein 1 (MSP1) expression (Fig. 6A and B). At 52 h pi, most of the WT parasites had formed and released merozoites or were in the final stages of merozoite formation (Fig. 6B) but, although the fabb/f− liver stages had somewhat increased in size in comparison with 44 h, there was still no significant expression of MSP1 (Fig. 6B). High-magnification differential interference contrast images of the liver stage parasites (Fig. 6C) showed that at 44 h pi, cytomere formation, which is due to multiple invaginations of the liver stage plasma membrane, was occurring in WT but not in the fabb/f− parasites and at 52 h pi, merozoites had formed and segregated in WT liver stages but not in the fabb/f− liver stages (Fig. 6C). The sporozoite infectivity and liver stage developmental data therefore suggest that fabb/f− parasites are normal with regard to hepatocyte invasion and liver stage development up to a point in late liver stage schizogony. However, fabb/f− liver stages do not reach maturity at the time point at which WT liver stages do and are unable to form infectious merozoites, explaining the lack of onset of blood stage infection from the liver. This is an unprecedented phenotype in late liver stage differentiation caused by the lack of an intrinsic metabolic pathway.

10To test whether the FAS II pathway might be dispensable for a clinically relevant human malaria parasite, we decided to study FabI in the NF54 strain of P. falciparum. We were able to delete the P. falciparum FabI gene from blood stage parasites using double homologous cross-over recombination in association with positive and negative selection. Deletion of the gene was confirmed by Southern blot analysis of the parental line (data not shown) and two clonal lines derived from the parental line (Fig. S4). We then compared the replication of the P. falciparum fabi− blood stages with that of the WT NF54 strain. We saw no significant differences in replication efficiency between WT and fabi− (Fig. 7 and data not shown) in the growth assay, which strongly suggests that, as for P. yoelii, the FAS II pathway is dispensable for P. falciparum blood stage growth.

11Although many metabolic functions have been assigned to the Plasmodium apicoplast, including FAS II, it is not known if these pathways are essential for every life cycle stage of the malaria parasite. As the parasite inhabits both extracellular and intracellular niches and replicates in the mosquito vector and mammalian host, its ability to scavenge host nutrients to support replication presumably varies greatly depending on its environment. The study presented herein analyses for the first time the importance of a metabolic pathway throughout the parasite life cycle. We demonstrated that apicoplast-targeted FAS II is only necessary for Plasmodium late liver stage development. Thus, in all other life cycle stages either the parasite synthesizes fatty acids by a yet unidentified de novo pathway or the parasite is able to scavenge all the fatty acids it requires from the host. Moreover, FAS II is only needed late in liver stage schizogony. We have previously shown that the liver stage PV membrane-resident protein UIS3 interacts with the hepatocyte lipid carrier liver-fatty acid binding protein. It is possible that this interaction allows for the transfer of lipids from the hepatocyte cytosol to UIS3 and subsequently to the developing liver stage and this concept has recently been buoyed by the fact that P. falciparum UIS3 cocrystallizes with the lipid phosphatidylethanolamine. Thus, the parasite liver stage might have developed a means of directly transferring lipids from its host. Nevertheless, host lipids alone are clearly not sufficient for completion of liver stage development. In the malaria model under study, no difference was seen in the first half of liver stage development (up to 24 h pi) between WT and fabb/f− parasites and it was only after 24 h that the fabb/f− liver stages showed growth retardation, accompanied by lack of cytomere formation and subsequent merozoite differentiation. Thus, during the complete P. yoelii life cycle, FAS II is required only for the final stages of parasite transition from its first site of infection in the liver to the blood. We do not currently know why FAS II is only necessary for late liver stage development. It is possible that the sheer amount of membrane biogenesis required for the formation of tens of thousands of merozoites cannot be met by host lipid scavenging and thus also relies on parasite-derived fatty acid synthesis to give a final boost to the formation of the merozoite membrane phospholipid bilayers. Alternatively, FAS II could be necessary to provide a particular fatty acid that is necessary for late liver stage development. In the sleeping sickness parasite, Trypanosoma brucei, the bloodstream form evades the host's immune response by expressing continually switching variant surface glycoprotein molecules. The variant surface glycoprotein is attached to the plasma membrane by a glycosylphosphatidylinositol (GPI) anchor whose fatty acids are exclusively myristate. The continuous supply of myristate cannot be met by the bloodstream and T. brucei has a unique de novo fatty acid synthesis pathway to supply its myristate needs. Perhaps Plasmodium FAS II is fulfilling a similar need by supplying particular fatty acids necessary for late liver stage schizogony that the parasite cannot obtain from its host. It is of interest to note that MSP1 is a GPI-anchored protein. However, we currently do not know if GPI biosynthesis is abrogated in FAS II-deficient liver stages. Notwithstanding, we observed lack of MSP1 in liver stages of FAS II knockout parasites, suggesting that MSP1 could play an essential role in the formation of exo-erythrocytic merozoites.

12Our data show that some FAS II enzymes are initially expressed in the salivary gland sporozoite based on both epitope tagging of FabI and our qRT-PCR data. This also revealed a high transcript abundance for FabI and FabB/F in salivary gland sporozoites although this was not the case for FabG and FabZ. We currently do not understand the functional significance of these increases; nevertheless, the pathway is clearly not necessary in this life cycle stage as both fabb/f− and fabz− sporozoites infected the mosquito salivary glands and were able to initiate liver stage infection. Previously, it was shown that depletion of UIS3 and UIS4, proteins that are initially expressed in sporozoites and localize to the PV membrane during liver stage development, as well as the sporozoite proteins P52 and P36, and the sporozoite asparagine-rich protein 1, cause arrest early in liver stage development. However, depletion of these proteins, as shown here for FabB/F and FabZ, does also not affect salivary gland sporozoite maturation. Furthermore, the liver stage phenotype for fabb/f− parasites is unique, when compared with the above as the parasites do not arrest early in liver stage development but grow substantially and undergo schizogony. At 52 h pi, the fabb/f− liver stages still seemed viable, based on the presence of an intact PV membrane (visualized by Hep17 expression). Nevertheless, liver stage development is not completed and the parasites might be cleared by host defence mechanisms although this requires further investigation.

13Our studies have also shown that FAS II is not necessary for either the mosquito stage or blood stage of the P. yoelii life cycle. It is currently not known how the parasite utilizes host lipids whilst developing on the mosquito midgut but it is well documented that host serum fatty acids are utilized by parasite blood stages for growth. Our results appear to contradict previous work demonstrating that triclosan, an inhibitor of FabI, is able to kill cultured P. falciparum blood stages and in vivo rodent blood stage infections of P. berghei. Others have also concluded that FAS II inhibitors are directly interacting with their apicoplast targets in Plasmodium blood stages, thereby inhibiting parasite growth. However, it is possible that the FAS II inhibitors used are having off target effects on parasite growth. This has previously been shown in T. brucei, where it was concluded that triclosan killing may be due to a non-specific perturbation of subcellular membrane structure leading to dysfunction in sensitive membrane-resident biochemical pathways. Furthermore, studies of the effect of triclosan on several microorganisms have concluded that the interaction of triclosan with the bacterial cell is complex and its lethality cannot be explained solely by the inhibition of metabolic pathways such as FAS II. Thus, triclosan could be killing P. falciparum blood stages by inhibiting a vital process other than FAS II. It has recently been shown based on transcriptional data obtained form malaria patient isolates, that there appears to be three distinct P. falciparum blood stage physiological states. These three states closely resemble (i) active growth; (ii) starvation; and (iii) environmental stress. In the starvation state, the authors noted an upregulation of FAS II genes when compared with active growth. These data suggest that under active glycolytic growth conditions, which were similar to the P. falciparum 3D7 growth conditions in vitro for which transcriptome profiles have been previously published, FAS II is not significantly involved in blood stage replication.

14This study has concentrated on the effect of FAS II depletion on liver stage development in the rodent malaria parasite P. yoelii. However, we have also shown that the deletion of FabI from the human malaria parasite, P. falciparum has no apparent effect on blood stage replication when compared with WT parasites. This result, which converges on the results we obtained for the deletions of P. yoelii FabB/F and FabZ, demonstrates that FAS II is not required in blood stages. Future work might determine whether deletion of FabI affects P. falciparum sporozoite infectivity in humans; however, this requires a clinical investigation. Collectively, the data suggest that the metabolic pathways present in the Plasmodium apicoplast are not always necessary to support parasite progression through the specific parts of the life cycle. Thus, this unique organelle is likely to perform its critical metabolic functions at only certain time points during the Plasmodium life cycle and the functions it performs will be directly related to the needs the parasite cannot fulfil by nutrient uptake from the host. The Plasmodium apicoplast, as well as being the centre for FAS II, is also thought to harbour the only pyruvate dehydrogenase complex the parasite possesses. It is possible that pyruvate dehydrogenase is solely required by the apicoplast for the formation of acetyl CoA which is subsequently utilized by FAS II. Further studies are needed to address this issue.

15Although our findings concerning liver stage development were generated with the rodent malaria parasite P. yoelii, the high conservation of FAS II among Plasmodium species suggests that FAS II is also essential for P. falciparum liver stage development. This might have consequences for the direction of antimalaria FAS II inhibitor drug development. Rather than concentrating on the P. falciparum blood stage, research into FAS II inhibitors should concentrate on their efficacy against the initial, clinically silent liver stage of infection. This might significantly contribute to the goal of eradicating malaria.

16Six- to eight-week-old female Swiss Webster (SW) mice and female BALB/c mice were purchased from Harlan (Indianapolis, IN). Animal handling was conducted according to institutional animal care and use committee-approved protocols.

17Plasmodium yoelii (17XNL) liver stage-infected hepatocytes were isolated at four time points post infection from infected mice: 12 h (LS-12), 24 h (LS-24), 40 h (LS-40) and 50 h (LS-50). Sporozoites were isolated from mosquito salivary glands at day 15 after infectious blood meal. Contaminating mosquito tissue was removed from sporozoite preparation by passing the extract over a DEAE cellulose column. For the preparation of parasites in the mixed blood stages, blood was harvested from infected SW mice when parasitaemia was at 5–10%. Lymphocytes were removed by passing the infected blood through a sephadex column. Purified blood stage schizonts were prepared from Nycodenz purification of P. yoelii infected blood cultured for 12 h.

18Total RNA from each sample was extracted using Trizol (Invitrogen) and DNase treated using Turbo-DNA free (Ambion). Total RNA was then subjected to two rounds of linear amplification using the Amino Allyl Message Amp II aRNA Amplification Kit (Ambion) according to manufacturer's directions. First-strand cDNA was synthesized from 500 ng of amplified RNA (aRNA) using the Superscript III Platinum RT kit (Invitrogen). The resulting cDNA was diluted 1:5 with nuclease-free water. Primers (Table S2) were designed using Primer Express v3.0 (Applied Biosystems). Designs were based on the mRNA sequence of the genes available at PlasmoDB. Amplicons were set to be between 100 and 200 bp. Real-time PCR analysis was performed on ABI prism 7300 Sequence Detection Systems using the SYBR Green PCR Master Mix (Applied Biosystems). The PCR reaction consisted of 12.5 μl of SYBR Green PCR Master Mix, 20 pmole of forward and reverse primers and 5 μl of diluted cDNA in a total volume of 25 μl. PCR cycling conditions were performed using the default conditions of the ABI Prism 7300 SDS Software.

19Quantification of gene expression was done using the Relative Standard Curve Method (Applied Biosystems bulletin). The standard is prepared from a mixture of aRNA (salivary gland sporozoites, blood stages and liver stages) in a 1:1:1 ratio. First-strand cDNA is prepared from the standard and dilutions of 1:1, 1:5, 1:10, 1:20 and 1:50 of the resulting cDNA was used as templates for real-time PCR for each primer pair. The relative quantity of gene in the cDNAs from the seven aRNA samples (salivary gland sporozoites, LS-12, LS-24, LS-40, LS-50, mixed blood stages and blood stage schizonts) is interpolated from the corresponding standard curve. Expression of the FAS II genes was normalized to the expression of three housekeeping genes: P. yoelii 18S, 14-3-3 protein (PY01841) and the mitochondrial EF-TU gene (PY06134) in each cDNA sample. Normalized quantity of each target gene is expressed as the ratio of the relative amount of target gene over the average quantity of the three housekeeping genes.

20To epitope tag FabI, a quadruple (4×) myc tag sequence followed by a stop codon was introduced into the b3D.DT^H.^D vector (Catalog # MRA-80 in the MR4 Malaria Research and Reference Reagent Resource Center; http://www.mr4.org). A graphic representation of the construction of the plasmid and its subsequent transfection into P. yoelii blood stages is given in Fig. S1. Approximately one kilobase pairs (kb) of the 3′ untranslated region of the P. berghei DHFR/TS gene was added to the C-terminus of the 4× myc tag to ensure stability of the recombinant messenger RNA. The P. yoelii FabI gene (PY03846), including approximately 1 kb of sequence upstream of the start codon was amplified from P. yoelii 17XNL genomic DNA and cloned in frame (without the stop codon) and upstream of the 4× myc tag. The resulting plasmid was linearized with BsaI for integration into P. yoelii 17XNL blood stage schizonts using standard procedures. Integration of the plasmid to create PyFabI-myc gave rise to a parasite line that expressed two copies of FabI, both with the endogenous promoter and one containing the 4× myc epitope tag. Oligonucleotide primers used are in Table S2. A similar strategy was used to generate PyFabZ-myc and PyFabG-myc.

21In vitro assays were conducted using the human hepatoma cell line HepG2 expressing the tetraspanin CD81 (HepG2:CD81) cultured in Dulbecco's modified Eagle's medium with 10% fetal calf serum at 37°C and 5% CO2. Infections were done by adding 5 × 104 sporozoites to individual chambers of an 8 well chamber slide (Laboratory-Tek Permanox eight-well chamber slide; Nalge Nunc International, Rochester, NY) which had been seeded with 105 subconfluent HepG2:CD81 cells the previous day. The slide was then centrifuged at 500 g for 2.5 min to aid sporozoite infection. Sporozoites which had failed to invade cells were removed after 2 h and the media were replaced. For the liver stage development assay, infections were maintained for various time periods after the addition of salivary gland sporozoites. Subsequently the infected cells were fixed in 4% paraformaldehyde in phosphate buffered saline (PBS) for 10 min, then blocked and permeabilized in PBS with 2% bovine serum albumin and 0.2% Triton-X 100 (PBS/BSA/Triton) for IFA. IFA was carried out in PBS/BSA/Triton. The double staining was performed using a mouse monoclonal anti-P. berghei HSP70 primary antibody, a parasite cytoplasmic marker and a rabbit polyclonal anti-myc antibody (Santa Cruz Biotechnology, Santa Cruz, CA), which recognizes the recombinant apicoplast-expressed FabI-myc. Fluorescent staining was achieved with Alexa Fluor-conjugated secondary antibodies (Invitrogen Corporation, Carlsbad, CA) specific to rabbit (Alexa Fluor 488, green) and mouse (Alexa Fluor 594, red) IgG. Cells were stained with 4′,6′-diamidino-2-phenylindole (DAPI) to visualize the DNA and mounted with FluoroGuard anti-fade reagent (Bio-Rad, Hercules, CA). Preparations were analysed using a fluorescence inverted microscope (Eclipse TE2000-E; Nikon), and images were acquired using Olympus 1 × 70 Delta Vision deconvolution microscopy.

22For the targeted deletion of the FabB/F (PY04452) and FabZ (PY01586) genomic loci, two DNA fragments containing approximately 0.8 kb of the 5′UTR and 3′UTR of the gene were amplified using P. yoelii 17XNL genomic DNA as a template. The two fragments were cloned into the b3D.DT^H.^D targeting vector between the T. gondii DHFR/TS gene, which allows selection of recombination events with pyrimethamine. The plasmid was transfected into P. yoelii 17XNL blood stage schizonts using standard procedures. Two independent clones of fabb/f− and fabz− parasites were obtained by limited dilution of parentals from independent transfection experiments. A graphic representation of the construction of the plasmid and its subsequent transfection into P. yoelii blood stages is given in Fig. 3 and the primers used in Table S2.

23To assay growth of non-lethal P. yoelii 17XNL WT, fabb/f− and fabz− blood stages, blood was removed from infected SW mice when parasitaemia was between 0.5% and 1.5%. The blood was diluted in RPMI-1640 media (HyClone, Logan, UT) so that 100 μl contained either 103 or 106 parasites. SW mice (four in each group) were then injected iv with 103 or 106 parasites (WT or fabb/f−). Percentage parasitaemia was followed as often as daily until clearance, by assay of Giemsa-stained blood smear.

24Anopheles stephensi mosquitoes were infected with P. yoelii WT, fabb/f− or fabz− parasites by blood feeding for 6 min on the first and second day using infected SW mice and subsequently maintained under a cycle of 12.5 h light/11.5 h dark and 70% humidity at 24.5°C. Gametocyte exflagellation capacity was evaluated microscopically before mosquito blood meal. Infected mosquitoes were dissected (at least 20 mosquitoes for each dissection) at days 10 and 14 (after the first infectious blood meal) to determine the presence of midgut oocyst sporozoites and the numbers of salivary gland sporozoites respectively.

25To analyse in vivo sporozoite infection and liver stage development, BALB/c mice were injected iv with 106 WT or fabb/f− sporozoites. For each parasite population, the livers were harvested from euthanized mice at several time points post infection (12, 24, 44 and 52 h). Livers were perfused with PBS, washed extensively with PBS and then fixed in 4% paraformaldehyde. Liver lobes were cut into 50 μm sections using a Vibratome apparatus (Ted Pella Inc., Redding, CA). For IFA, sections were permeabilized in Tris buffered saline (TBS) containing 3% H2O2 and 0.25% Triton X-100 for 30 min at room temperature. Sections were then blocked in TBS containing 5% dried milk (TBS-M) at least 1 h and incubated with primary antibody in TBS-M at 4°C overnight. Primary antibodies used were mouse monoclonal anti-circumsporozoite protein, mouse monoclonal anti-Hep17 and rabbit polyclonal anti-MSP1. After washing in TBS, secondary antibody was added in TBS-M for 2 h at room temperature in a similar manner as above. After further washing, the section was incubated in 0.06% KMnO4 for 10 min to quench background fluorescence. The section was then washed with TBS and cells were stained with DAPI to visualize the DNA and mounted with FluoroGuard anti-fade reagent (Bio-Rad, Hercules, CA). Preparations were analysed as above for fluorescence with the addition of acquisition of a differential interference contrast image.

26To compare the sizes of parasite liver stages at 44 h pi, liver sections labelled with Hep17 (see above) were sequentially scanned using Nikon fluorescence microscopy. The greatest diameter for each liver stage detected was determined by adjusting the z plane of the liver section and the area of the parasite was subsequently determined. For the WT, 251 liver stages were assayed and for the fabb/f−, 147. All parasite segments were less than 3000 μm2 and the total numbers of parasites were divided into quartiles by area.

27Targeting sequences 5′ and 3′ to P. falciparum FabI (PFF0730c) were cloned into plasmid pCC1 to facilitate positive–negative selection (Fig. S4). Restriction sites in the multiple cloning site were SacII/SpeI for the 5′ flank and AvrII/EcoRI for the 3′ flank. Sequencing was performed to confirm inserts and primers used are detailed in Table S2. Plasmid DNA was extracted by maxi prep kit (Qiagen). The NF54 line P. falciparum parasites were synchronized at ring stage with sorbitol 2 days prior to transfection. On the following day trophozoites were selected for WT cytoadherence properties by incubation in RPMI plus Gelofusine (Braun). Transfection of P. falciparum ring stages with 100 μg of DNA was performed by electroporation at 0.31 kV and 950 μF with a Bio-Rad Gene Pulser (Bio-Rad, La Jolla, CA). Cultures were placed on the positive selection drug WR99210 (Jacobus Pharmaceuticals, Princeton, NJ) 6 h post transfection and maintained as described. This was followed by negative selection against the cytosine deaminase/uracil phosphoribosyl transferase gene product with 5-fluorocytosine in order to obtain a parental line with double cross-over homologous recombination, which results in specific FabI gene deletion.

28Two individual clones with a FabI deletion were isolated (E6 and G8) and genotypic analysis was confirmed by Southern Blot (Fig. S4). Genomic DNA from WT NF54 and knockout lines was digested for 2–16 h with the following enzymes: 5′ test: KpnI/BglII and 3′ test: BglII/BamHI. Digested DNA was run on a 1% TAE agarose gel at 15 V for 18 h and transferred to Hybond-N membrane (Amersham) overnight at room temperature, UV cross-linked and pre-hybridized with herring sperm DNA for 2.5 h. A digoxygenin-labelled probe was prepared by PCR per supplier protocol (Roche) using the cloning primers. Hybridization was carried out for 18 h at 55°C. The blot was exposed to film for 10–60 min and developed per standard protocol.

29Parasite cultures of both the WT NF54 and the fabi− clone E6 containing mainly ring stages were synchronized twice within 4 h using sorbitol. Parasite density was determined and the culture was diluted to 0.8% parasitaemia, 5% haematocrit. Cultures were maintained under an atmosphere of reduced oxygen at 37°C and medium was refreshed every 24 h. Growth was monitored by Giemsa-stained thin blood smears every 24 h and for each determination of percentage parasitaemia, the number of infected erythrocytes per at least 2000 erythrocytes was recorded.