The Fatty Acid Biosynthesis Enzyme FabI Plays a Key Role In the Development of Liver Stage Malarial Parasites

Triclosan Activity Against Plasmodium Asexual Blood Stages Does Not Correlate With its Inhibition of Purified Recombinant FabI

We initiated a structure-guided medicinal chemistry program to improve the potency of triclosan by modifying substituents around its diaryl ether scaffold. This led us to synthesize 80 analogs (Figure 1A), for which only a single position was modified to reduce the likelihood of creating off-target activity. These analogs, grouped by the carbon position that was modified (see Figure 1B inset; detailed in Table S1 in the Supplemental Data available with this article online), were evaluated for their inhibition of cultured P. falciparum asexual blood stage parasites (3D7 and Dd2 lines), and separately, for inhibition of purified PfFabI enzyme. Activities were compared against triclosan, which yielded mean IC₅₀ values of 1.8 μM and 2.1 μM against the 3D7 and Dd2 lines and an EC₅₀ value of 73 nM against purified PfFabI. Individual analog series showed substantial differences in their inhibitory activities (Figure 1A). The 2′-position analog series afforded the most potent inhibitors in the parasite assays, yet showed minimal inhibition of PfFabI. Modifications at the 4′- or 5- position afforded a few modest improvements in efficacy, mostly against PfFabI, whereas changes at the 4- or 6- position generally produced less potent inhibitors. We also noted that compound series with similar mean potencies against PfFabI (e.g. the 4′- and 5- position analogs) exhibited nearly 10–fold differences in their mean antiparasitic activities. Further analysis revealed a lack of a significant association between enzyme and parasite inhibition, as evidenced by the Pearson r² values of 0.13 and 0.15 obtained by plotting these data for 3D7 and Dd2 respectively (Figure 1B; data not shown). While chemical properties such as membrane permeability and solubility might obscure close whole-cell and enzyme correlations for on-target compounds, our data nonetheless raised doubts that triclosan analogs acted against asexual blood stage parasites by inhibiting PfFabI.

In parallel with these studies, we re-evaluated the in vivo efficacy of triclosan. A previous report had documented that four days of subcutaneous injections of triclosan at 3 mg/kg/day could suppress P. berghei parasitemia by 75%, and that injections with doses of 38 mg/kg/day cured the infected mice with 100% efficacy (Surolia and Surolia, 2001). We tested both oral (PO) and subcutaneous (SC) routes of triclosan, administered over a range of 16 to 512 mg/kg/day, for three days with a twice-daily divided dose, to mice infected three days prior with P. berghei (KBG-173 line). Parasitemias were recorded on day 6 post-infection (i.e. one day after the last dose of triclosan) and were 61% or 57% (for PO or SC respectively) in control (infected and placebo-treated) mice. Increasing the triclosan concentrations from 16 to 128 mg/kg/day caused a dose-dependent decrease in parasitemias, to a minimum of 27% or 13% with 128 mg/kg/day triclosan administered PO or SC respectively (Figure 1C). Higher doses failed to further suppress the parasitemias.

Assessment of survival showed that all control mice died by days 10 (PO) or 12 (SC) (Figure 1D, E). Oral administration of 64 mg/kg/day triclosan yielded a slight extension in survival times, and doses of 128 or 256 mg/kg/day produced 30% survival rates measured at day 31. Increasing the dose to 512 mg/kg/day led to some early mortality and decreased overall survival rates, suggesting some toxicity. Via the SC route, triclosan was moderately more effective, although survival never exceeded 50%. In vivo tests with several analogs (compounds 18, 20, 22, 41, 45 & 60 in Table S1), whose in vitro potencies were comparable to triclosan, revealed no improvements over the parent compound (A. Ager and D. Jacobus, unpublished data). We concluded that, under our experimental conditions, triclosan had reduced antimalarial potency in vivo as compared to the earlier report (Surolia and Surolia, 2001). Furthermore, the in vitro potency of triclosan analogs did not correlate with their inhibition of FabI enzymatic activity.

Transgene Expression of a Mutant FabI That is Biochemically Resistant to Triclosan Does Not Decrease P. falciparum Susceptibility to This Agent

To further investigate the role of FabI in the mode of action of triclosan, we transfected P. falciparum asexual blood stage parasites with plasmids expressing V5 epitope-tagged forms of pffabI that were either wild type (WT) or that encoded the A217V mutation. This mutation was selected because it confers a 7,000-fold decrease in triclosan binding affinity for recombinant PfFabI (Kapoor et al., 2004). To express these pffabI transgenes, we selected the calmodulin (PF14_0323) promoter, which is highly active in asexual blood stages (www.PlasmoDB.org). Integration of these plasmids (named pffabI(A217V)-V5-attP and pffabI(WT)-V5-attP; Table 1) into the P. falciparum genome was achieved using the Bxb1 serine integrase-mediated attBxattP system of recombination, which delivers transgenes into the attB-marked cg6 locus and results in a genetically and phenotypically homogeneous population of recombinant parasite lines (Nkrumah et al., 2006). The transfections, performed with Dd2^attB and 3D7^attB parasites, produced the transgenic lines PffabI(A217V)^Dd2, PffabI(WT)^Dd2, PffabI(A217V)^3D7 and PffabI(WT)^3D7 (Table 1; Figure S1A). Southern blot analysis confirmed correct integration of the pffabI transgenic copies into the cg6-attB site and the predicted organization of the endogenous pffabI locus (Figure S1B; data not shown). Western blot analysis with anti-V5 antibodies revealed the expression of ~46 kDa V5-tagged PfFabI proteins in all four lines (Figure S1C). Immunofluorescence assays with the PffabI(A217V)^Dd2 line showed co-localization of PfFabI-V5 and the apicoplast-resident acyl carrier protein (ACP) in a compartment distinct from the nucleus (stained with Hoechst 33342) and the mitochondrion (stained with MitoTracker Red), thus confirming trafficking of V5-tagged PfFabI to the apicoplast (Figure 2A, B). We proceeded to measure the triclosan susceptibility of our transgenic and parental lines. Results from drug susceptibility assays showed no significant difference in either IC₅₀ or IC₉₀ values between recombinant lines expressing mutant or WT PfFabI (Figure 2C; values provided in legend).

P. falciparum FabI is Not Expressed at Detectable Levels in Asexual Blood Stage Parasites

Subsequent Northern blot experiments detected the presence of pffabI transcripts only in the transgenic lines (PffabI(A217V)^Dd2, PffabI(WT)^Dd2, PffabI(A217V)^3D7 and PffabI(WT)^3D7) that express an additional pffabI copy from the highly active mature stage calmodulin promoter, and not in the lines expressing endogenous pffabI alone (Dd2^attB, Dd2, 3D7^attB and 3D7; Figure 2D). We attribute the apparent increase in pffabI transcripts in the PffabI(A217V)^{Dd2 and} PffabI(A217V)^3D7 lines, compared to the PffabI(WT)^Dd2 and PffabI(WT)^3D7 lines, to the higher proportions of mature stage parasites in the former at the time of RNA harvest. To confirm the lack of endogenous pffabI expression at the protein level, we raised rabbit polyclonal antiserum against purified recombinant PfFabI. Western blot analysis with extracts of parasites expressing calmodulin promoter-driven PfFabI-V5 showed that the antiserum and monoclonal antibodies against the V5 epitope tag both detected a ~46 kDa protein (Figure 2E). The anti-PfFabI antiserum did not detect any protein in extracts of control parasites expressing pffabI from its endogenous promoter (Figure 2F).

FabI is Not Required for Normal Propagation of P. falciparum Asexual Blood Stage Parasites and its Absence Does Not Alter Parasite Susceptibility to Triclosan

These findings led us to question whether pffabI is required for P. falciparum asexual blood stage growth in vitro. To test this, we designed a DNA construct (pcam-bsd-ΔpffabI), which contained an internal region of the pffabI coding sequence (encoding amino acids 98 to 295) such that homologous recombination between this fragment and the endogenous pffabI gene would separate the full-length sequence into two truncated fragments (Figure S2A). The upstream fragment lacked the 3′ end of the gene corresponding to amino acids 296–432 (thereby eliminating the β6–β9 helices and β6–β7 strands that contribute to forming the NADH-binding Rossman fold). The downstream fragment lacked a promoter and the first 98 amino acids that included the bipartite targeting sequence predicted to be required for protein trafficking to the apicoplast (Perozzo et al., 2002). Transfection of cultured Dd2 parasites with this knockout construct resulted in the generation of parasite clones (PfΔfabI¹ and PfΔfabI²) in which the pffabI gene had been disrupted by single site crossover and plasmid integration, as confirmed by PCR and Southern blot analyses (Table 1; Figure S2B, C). Measurements of parasitemia over a two-month period revealed equivalent growth rates (averaging 5.0 to 5.6–fold multiplication per 48 hr cycle of RBC invasion, intracellular development and egress) between these knockout clones and parental Dd2. These data demonstrated the non-essentiality of pffabI for asexual blood stage propagation, and implied that the activity of triclosan against these stages could not result from inhibition of PfFabI. This was confirmed with drug susceptibility assays that revealed similar triclosan susceptibilities in PfΔfabI¹, PfΔfabI² and the parental Dd2 line (mean ± SEM IC₅₀ values of 2.2±0.2, 2.4±0.3 and 2.1±0.3 μM respectively, derived from three separate experiments performed in duplicate).

Deletion of P. berghei fabI, the pffabI Ortholog, Does Not Affect Propagation of Blood Stage Parasites In Vivo

Our P. falciparum in vitro data led us to examine whether this protein was essential for proliferation of asexual blood stages in vivo. For this, we used the highly virulent P. berghei ANKA rodent malaria model. Comparisons of the amino acid sequences of PfFabI and its predicted ortholog in P. berghei, PbFabI (PB000088.02.0), revealed 62% identity and 74% similarity (Figure S3A). Bacterial expression and purification of PbFabI enabled us to elucidate its structure at 2.5Å resolution. Superimposing this with the known PfFabI structure (Perozzo et al., 2002) revealed a nearly identical organization with each subunit in the tetramer containing 9 α-helices and 7 β-sheets (Figure S3B). Detailed inspection of the active sites revealed these to be indistinguishable (Figure S3C). From these studies, we can confidently predict that PbFabI and PfFabI fulfill the same enzymatic function for Plasmodium parasites.

We transfected P. berghei ANKA parasites with a DNA construct termed pLitmus28-ΔpbfabI. This was designed to permit double crossover, resulting in complete deletion of the endogenous pbfabI locus and its replacement by the T. gondii dihydrofolate reductase-thymidylate synthase (Tgdhfr-ts) selectable marker that confers resistance to pyrimethamine. From this transfection, we selected mutant parasites and used limiting dilution to obtain a clone, termed PbΔfabI (Table 1). PCR and Southern blot analyses confirmed correct integration of the DNA construct and deletion of the pbfabI coding sequence in this clone (Figure S4A, C, D). As a “knock-in” control, we utilized a similar double crossover strategy to replace the endogenous gene with a construct that reinserted a full-length, functional pbfabI gene under control of the endogenous promoter, as well as the Tgdhfr-ts selectable marker. This yielded the PbfabI^Rec clone, whose recombinant locus was confirmed by PCR and Southern blot hybridization (Figure S4B, C, D). Measurements of parasitemia in mice infected with PbΔfabI, PbfabI^Rec or parental ANKA revealed similar rates of proliferation, calculated to be 4.8±1.4, 5.5±2.2 and 4.4±1.7 per 24 hr cycle respectively in two comparative experiments with groups of 4 mice each (values represents means±SD; Figure S4E). These values were not statistically different between lines, as determined by Mann-Whitney tests. Thus, there was no substantial decrease in asexual blood stage in vivo viability upon deletion of pbfabI.

Drug susceptibility assays with P. berghei asexual blood stages tested ex vivo showed equivalent triclosan IC₅₀ values in the PbΔfabI, PbfabI^Rec and ANKA lines (means±SEM of 1.4±0.1, 1.5±0.2 and 1.2±0.1 μM respectively). Control assays with the unrelated antimalarial chloroquine produced IC₅₀ values of 11.0±0.1, 13.0±0.6 and 10.4±0.3 nM in these lines respectively. These results, combined with the P. falciparum studies, conclusively demonstrate that the blood stage activity of triclosan is not attributable to inhibition of FabI.

Plasmodium Asexual Blood Stage Parasites Lacking FabI Can Produce FA Species

The availability of parasite lines lacking FabI allowed us to determine whether Plasmodium asexual blood stages utilize the FAS-II pathway to synthesize FA de novo. We incubated synchronized trophozoite stage P. falciparum and P. berghei control and fabI knockout parasites with [¹⁴C]-labeled acetate, a radiolabeled FA precursor, then extracted the free FA that had incorporated this substrate and analyzed them by reversed-phase High Performance Liquid Chromatography (HPLC). In P. falciparum, this led us to detect radiolabeled C₁₆ and C₁₈ FA in both the PfΔfabI¹ and Dd2 lines (Figure 3A, B). Thus, extension of FA could occur in the absence of the key FAS-II enzyme FabI. In a separate experiment, we incubated PfΔfabI¹ and other P. falciparum lines with [¹⁴C]-labeled acetate, extracted their FA and analyzed them by reversed-phase thin layer chromatography. This confirmed the production of radiolabeled C₁₆ and C₁₈ independently of FabI (Figure S2D). We note that these findings are in contrast with an earlier report that P. falciparum asexual blood stage parasites synthesize C₁₀ to C₁₄ FA (Surolia and Surolia, 2001).

[¹⁴C]-acetate incorporation studies with the P. berghei lines produced evidence of de novo FA elongation with PbΔfabI parasites, with no visible difference between the PbΔfabI and parental ANKA lines in terms of the lengths of FA that were produced (Figure 3C, D). In contrast to P. falciparum, the rodent parasites synthesized FA chain lengths of C₁₂ to C₂₄. No radiolabeled FA were observed with rodent or human uninfected RBC controls (data not shown). These data provide evidence that the two Plasmodium species differ in the range of FA that they can extend de novo; yet they share the common characteristic that FabI is not required.

P. berghei Sporozoites Lacking FabI Are Markedly Attenuated in Their Ability to Progress to Asexual Blood Stage Infections

The availability of a P. berghei line (PbΔfabI) lacking FabI made it possible to explore the role of this protein in other stages of the parasite life cycle. We observed that gametocyte production, gamete fertilization, and the subsequent development of ookinetes and oocysts appeared unaffected by the absence of FabI, as judged by the similar numbers of oocysts that developed in Anopheles stephensi mosquitoes fed on mice infected with the PbΔfabI, PbfabI^Rec or ANKA lines (data not shown). In two separate experiments, we observed no difference in the numbers of oocyst and salivary gland SPZ produced by these lines (data not shown).

To determine whether FabI plays a role in the infectivity of SPZ to the mammalian host, salivary gland SPZ were dissected and inoculated into the tail vein of C57BL/6 mice. This inbred strain of mouse was chosen as it is more susceptible than other inbred or outbred mouse strains to P. berghei SPZ infections (Scheller et al., 1994). Experiments were performed on three separate occasions and were highly reproducible. Intravenous injections of 1,000 ANKA or PbfabI^Rec SPZ produced patent blood stage infections that were microscopically detectable in 16/16 and 15/15 mice by day five (Table 2). In contrast, injection of 1,000 PbΔfabI SPZ produced a blood stage infection in only 5/16 mice, with the infected mice showing a delay in patency of 4 days. Increasing the PbΔfabI inoculum to 10,000 SPZ resulted in patent infections in 13/15 mice, with those mice again showing an average delay of 4 days compared to controls.

We also assayed the infectivity of SPZ delivered by mosquito bite, to test for any defect in SPZ tissue traversal and migration from the dermis to the liver (Silvie et al., 2008b). Groups of 20 infected mosquitoes were allowed to feed on each mouse, with 5 mice tested per P. berghei line. Based on earlier studies, we estimate that each infected mosquito intradermally delivers ~120 SPZ (Medica and Sinnis, 2005), yielding an approximate inoculum of 2,000 SPZ. Results showed that each mouse infected with ANKA or PbfabI^Rec parasites, as well as 4 of the 5 mice infected with PbΔfabI SPZ, developed a patent blood stage infection. However, the latter group showed a 5-day delay (Table 2).

Taken together, the data demonstrate that P. berghei parasites lacking FabI produce SPZ that are highly attenuated in their infectivity to the mammalian host. We note that all “breakthrough” PbΔfabI asexual blood stage infections became fulminant and lethal by days 20–29. This suggests that once PbΔfabI parasites developed into blood stages, they showed no loss of virulence compared to WT parasites. PCR analysis of breakthrough infections confirmed that they resulted from PbΔfabI parasites, and not from contamination with ANKA or PffabI^Rec parasites (Figure S4F).

PbΔfabI Sporozoites Typically Fail to Produce Infectious Mature Liver Stage Parasites

To investigate the cause of this decreased infectivity of PbΔfabI SPZ, we first examined cell traversal and invasion of hepatocytes. The former occurs when SPZ transit through cells prior to initiating liver stage development by forming a parasitophorous vacuole inside the invaded cell (Silvie et al., 2008b). In two independent experiments, rates of cell traversal were similar, with a mean of 9–13 dextran positive (i.e. traversed) cells per field. PbΔfabI and PbfabI^Rec lines were also found to be equally competent for invasion, with 34–38% success in invading Hepa 1–6 cells. We next assessed the maturing liver stage parasites. At 24, 36 or 48 hr post-invasion, PbΔfabI and PbfabI^Rec parasites stained with antibodies specific for the P. berghei cytosolic protein HSP70 (PB000817.02.0) showed equivalent numbers and developmental stages (data not shown). We also tested for fabI transcription in these stages. RT-PCR studies from infected Hepa 1–6 cells harvested 40 hr post-invasion revealed pbfabI mRNA transcripts in PbfabI^Rec but not PbΔfabI liver stage parasites (Figure S4G). These results indicate that fabI is normally transcribed by liver stage parasites, however the lack of expression in PbΔfabI parasites did not affect SPZ cell traversal, invasion of hepatocytes, or the initial stages of intra-hepatic development.

We proceeded to investigate later stages of liver stage maturation using the HepG2 hepatoma cell line, which is able to support SPZ invasion and liver stage development through to the production of merozoites that are infectious for RBC. Immunofluorescence assays (IFAs) with mature hepatic stages were performed with antibodies that recognize the P. berghei parasite proteins Exp1 (PB000484.01.0) or MSP1 (PB000172.01.0). Exp1 is expressed throughout trophozoite development and schizogony and is exported into the parasitophorous vacuolar membrane (PVM) that separates the parasite from the host cell cytosol. In late liver stages, the Exp1-positive PVM is typically observed as a closed circular structure around the multiplying parasite nuclei (Sturm et al., 2006). In schizont stages, MSP1 is expressed and becomes integrated into the parasite plasma membrane (PPM). The PPM invaginates around pockets of parasite material during the cytomere stage, and ultimately forms the membrane of individual merozoites (Sturm et al., 2008).

These antibodies revealed a striking difference between ANKA and PbΔfabI parasites very late in their liver stage development. At 60 hr post SPZ invasion of HepG2 cells, 59% of the ANKA parasites were found to have developed into an advanced parasite stage marked by MSP1-positive PPM invaginations (Figure 4A; Figure S5A). Many of these parasites were observed at the cytomere stage in which the PPM surrounded large groups of parasite nuclei (see row 2 in Figure S5A). The remaining 41% of parasites were MSP1-negative, indicating either delayed or aberrant development. In contrast, almost all PbΔfabI liver stage parasites (99.5%) were negative for MSP1 staining, despite having initiated their development inside an Exp1-positive PVM (Figure 4A). In addition to the lack of MSP1 expression in PbΔfabI parasites, nuclear division was clearly impeded, as evidenced by the limited number of DAPI-positive parasite nuclei (Figure 4A). The very few MSP1-positive PbΔfabI liver stage parasites that we did observe were restricted in size, with minimal PPM invaginations (Figure 4A; Figure S5A). In support of this, at 60–65 hr post-invasion we recorded no cytomere stage in over 3,500 PbΔfabI liver stage parasites, whereas cytomeres were observed in 266 ANKA parasites out of a total of 3,366. At these late stages of parasite development, ANKA parasites began to degrade the PVM, as evidenced by their lack of the typical Exp1-positive closed circular structures seen in earlier stages (Figure S5B). This resulted in the generation of large clusters of MSP1-positive merozoites that filled the entire hepatocyte cytoplasm. Of these PVM-degraded ANKA-infected cells, 87% contained MSP1-positive merozoites. The remaining 13% were MSP1-negative, suggesting that these had undergone aberrant development and had failed to produce viable merozoites (Figure S5B). In contrast, every PbΔfabI parasite that was found to have a non-intact PVM was MSP1-negative and was not producing mature merozoites (Figure S5B).

At the 65 hr time point, we also recorded the numbers of infected cells that had detached from the monolayer into the culture supernatant. These so-called “detached cells” harbor merozoites that have been released into the host cell cytoplasm following normal degradation of the PVM (Sturm et al., 2006). In three independent experiments, we did not observe a single detached cell with the PbΔfabI parasites. In contrast, detached cells numbered 210, 236 and 106 with ANKA parasites and 298, 136 and 103 with PbfabI^Rec parasites (Figure 4B). From these cultures, we also recorded the number of merosomes, i.e. the membrane-bound clusters of merozoites devoid of host nuclei that egress from the infected hepatocytes (Sturm et al., 2006; Sturm et al., 2008). Merosomes were never observed in PbΔfabI liver stage cultures in any of these three experiments. By comparison, ANKA parasites produced 69, 174 and 41 merosomes, while PbfabI^Rec parasites produced 77, 68 and 13 merosomes respectively (Figure 4B). Similar results were obtained in three additional experiments that examined liver stage parasites 70 hr post-invasion (data not shown), confirming a key role for FabI during the final maturation of the liver stage schizont and the formation of merozoites.

While we never observed PbΔfabI merosomes, even after an extended culture period of 90 hr, we nevertheless recorded rare instances of MSP1-positive PbΔfabI merozoites. We also observed one instance of a cytomere stage at 70 hr post-invasion. The formation of these few merozoites and their passage into the blood stream might account for the PbΔfabI breakthrough blood stage infections observed in vivo.

Parasite Propagation

P. falciparum lines (Table 1) and P. berghei ANKA parasites were propagated as described in the Supplemental Data. All experiments involving rodents were conducted in fully accredited animal facilities and were approved by the Institutional Animal Care and Use Committees of Columbia University, the Albert Einstein College of Medicine, the New York University Medical Center, the University of Miami and the Bernhard Nocht Institute for Tropical Medicine.

Parasite in vitro and in vivo Drug Susceptibility Assays

The synthesis of triclosan analogs has been previously described ((Freundlich et al., 2007) and references therein). For enzyme inhibition assays, the reaction mixtures contained 50 nM PfFabI, 400 μM NADH and 40 μM NAD⁺, and were initiated with 300 μM butyryl-CoA. Inhibition of PfFabI–mediated butyryl-CoA reduction was assessed spectrophotometrically by measuring the oxidation of NADH to NAD⁺ at 340 nm (Freundlich et al., 2007). EC₅₀ values represent the analog concentration that inhibited maximal PfFabI activity by 50%. Inhibition of P. falciparum in vitro or P. berghei ex vivo parasite growth was measured using [³H]-hypoxanthine assays, and the IC₅₀ values calculated using linear regression (see Supplemental Data). All compounds were tested in duplicate on at least two separate occasions. For in vivo efficacy studies, CD-1 mice were infected intraperitoneally on day 0 with 5 × 10⁴ P. berghei asexual blood stage parasites. Triclosan (Vita-Pharm, Valhalla, NY) was administered on days 3, 4 and 5 after infection, as two divided doses daily spaced 6 hr apart delivered either PO or SC. Parasitemias were determined from Giemsa-stained smears of tail blood, collected on day 6 and twice a week thereafter until day 31 (see Supplemental Data).

Plasmid Constructs, Parasite Transfections, Nucleic Acid and Protein Analyses, Immunofluorescence Assays, and Structural Elucidation of PbFabI

These are detailed in the Supplemental Data. Primers are listed in Table S2 and transfection plasmids and parasite lines are listed in Table 1.

FA extraction and HPLC analysis

P. falciparum- or P. berghei-infected RBC were labeled with [¹⁴C]-acetate (10 μCi/ml) for 6 hr or 24 hr respectively at 37°C. Parasite pellets were obtained by saponin lysis, washed twice to remove host cell components, and the FA analyzed by reversed-phase HPLC (see Supplemental Data).

P. berghei Infection of Mosquitoes, Sporozoite Invasion and Cell Traversal Assays, and Analysis of Liver Stage Development

Experimental conditions are detailed in the Supplemental Data.

Determination of P. berghei Prepatent Periods in Mice

To determine the in vivo infectivity of mutant and control parasite lines, naïve C57BL/6 mice were injected intravenously with 1,000 or 10,000 P. berghei salivary gland SPZ, or subjected to the bite of 20 infected mosquitoes that were allowed to probe for 6 min. Asexual blood stage infection was determined by Giemsa-stained blood smears prepared on days 3 through 28 after SPZ inoculation.