Summary
Malaria is a human parasitic disease that is transmitted by a mosquito vector. Plasmodium parasites, the causative agents, differ in their infectivity and virulence to the mammalian host, but the mechanistic underpinnings of this variation remain unknown. As mosquitoes provide a nutrient-rich niche for development of transmissible stages, we examined the role of lipids in parasite development and infectivity by disrupting lipid trafficking in mosquito adults. We show that depleting the major mosquito lipoprotein lipophorin deprives parasites of neutral lipids, arrests oocysts growth and impairs sporozoite formation. Importantly, lipid deficiency decreases parasite mitochondrial membrane potential and severely compromises sporozoite infectivity and virulence in the mammalian host. Our findings demonstrate the requirement of mosquito lipids for Plasmodium metabolism, and uncover the mitochondrial contribution to parasite infectivity and virulence. By drawing a connection between vector nutrition and malaria virulence, our results redefine the paradigm of vector-host-pathogen interactions.
INTRODUCTION
Nutrition plays a key role in host-pathogen interactions, as parasites hijack host metabolites to survive and propagate. Plasmodium parasites, the causative agents of malaria, are no exception. Malaria is the deadliest infectious disease transmitted to humans by anopheline mosquitoes. Multiple reports have implicated larval diet and adult blood meals in determining parasite loads within a mosquito (Moller-Jacobs et al., 2014; Okech et al., 2004; Ponnudurai et al., 1989; Shapiro et al., 2016; Takken et al., 2013; Vantaux et al., 2016). Indeed, females produced by undernourished larvae are poor Plasmodium hosts (Moller-Jacobs et al., 2014; Shapiro et al., 2016; Takken et al., 2013; Vantaux et al., 2016), although additional blood feedings improve parasite survival (Okech et al., 2004; Ponnudurai et al., 1989). Together with proteins and carbohydrates, lipids constitute important building blocks of nutrition since they provide living organisms with energy, serve as the structural materials of cell and organelle membranes, and function as signaling molecules. In contrast to the defined role of lipids in Plasmodium development within a mammalian host (Grellier et al., 1991; Itoe et al., 2014; Lauer et al., 2000), the contributions of mosquito lipids remain unknown. Mosquitoes acquire lipids from their diet and store them in the fat body, an insect equivalent of mammalian liver. Lipid levels in the tissues are maintained by lipophorin (Lp), the hemolymph-circulating transporter that shuffles hydrophobic lipids between the fat body and energy demanding tissues (Atella et al., 2006; Van der Horst et al., 2009). The blood meal brings to the mosquito massive lipid resources that, in the course of the first days, are routed by Lp from the midgut to the developing ovaries for egg development and to the fat body for storage. Disrupting Lp function aborts ovary development and renders the parasites more vulnerable to the mosquito complement-like system (Mendes et al., 2008; Rono et al., 2010). Here we exploited Lp deficiency to examine the role of mosquito lipids in Plasmodium sporogonic development, transmission and virulence to the mammalian host.
RESULTS
Disruption of lipid trafficking between mosquito tissues mitigates Plasmodium sporogony
We first investigated whether Lp depletion disrupts lipid trafficking between mosquito tissues. To this end, females were injected with dsRNA against Lp and then fed on Plasmodium-infected blood. Lipid content of the tissues was gauged by Nile Red staining (Rudolf and Curcio, 2009). In control mosquitoes, blood feeding significantly increased levels of the Nile Red signal in the midgut and in the ovaries, but no change was observed in the fat body (Figure 1, A and B). While Lp depletion transiently increased lipid levels in the midgut, it abrogated the feeding-induced increase in lipid levels in the ovaries (Figure 1, A and B). Again, no changes in lipid levels were detected in the fat body. We conclude that blocking the blood meal-induced lipid trafficking by Lp depletion efficiently disrupts lipid transport between mosquito tissues.
We next examined the effects of disruption of lipid trafficking on Plasmodium development. To prevent large losses at early stages of parasite development inflicted by the mosquito complement-like system in the absence of Lp (Mendes et al., 2008; Rono et al., 2010), all infection experiments were performed with immune-deficient mosquitoes (Pompon and Levashina, 2015). When mosquitoes were infected with P. falciparum, a significant decrease in the size of mature oocysts was observed in Lp-depleted mosquitoes (Figure 1C). Strikingly, inhibition of lipid trafficking led to a 5-fold decrease in the number of salivary gland sporozoites (Figure 1D). Similar effects on sporogony and sporozoite loads were observed for P. berghei (Figure 1C and D). Further, the comparable decrease in numbers of midgut and salivary gland sporozoites suggested that Lp depletion inhibited parasite sporogony in the oocysts (Table S1). Taken together, these results show that Lp-mediated lipid trafficking is essential for sporogony of human and rodent Plasmodium parasites.
Lp-transported lipids are essential for Plasmodium sporogonic development
To investigate how disruption of lipid trafficking affected parasite sporogony, we followed development of P. berghei GFP-expressing oocysts by fluorescence microscopy (Franke-Fayard et al., 2004). A significant decrease in oocyst size was observed in Lp-deficient mosquitoes 7 days post infection (dpi) (Figure 2A), indicating that Lp deficiency arrests parasite growth at the onset of sporogony. We next examined whether oocysts accumulated lipids and the role of Lp in this process. A clear Nile Red signal was observed in the peripheral cytoplasm and vesicles (Figure 2B). Importantly, this signal was absent in the oocysts of Lp-depleted mosquitoes (Figure 2B and Figure S1A), suggesting that Lp is required for parasite provision with mosquito lipids. To investigate whether Lp was taken up by oocysts together with its lipid cargo, we used antibodies directed against the protein moiety of the Lp particle (Rono et al., 2010). In controls, The number of Lp-positive oocysts increased over time (25% and 72% at 7 and 13 dpi, respectively, Figure 2A). Lp signal was observed on the surface but not inside the oocysts (Figure 2C). As expected, no signal was detected in Lp-depleted mosquitoes (Figure 2C). Progressive accumulation of Lp on the parasite surface suggests that Plasmodium exploits Lp particles for lipid delivery to the oocysts rather than for uptake. Transmission electron microscopy analysis further confirmed that Lp depletion did not impact development of young oocysts (7 dpi) which displayed normal ultrastructure (Figure 2D and E). Instead, in Lp-depleted mosquitoes, almost half of the mature oocysts displayed cytoplasmic vacuolization and lack of classical membrane retractions characteristic of initiation of sporulation (Figure 2F). Moreover, these aberrant oocysts were surrounded by the electron-dense actin zones associated with dead ookinetes (Shiao et al., 2006) (Figure 2E and Figure S2). Collectively, these observations indicate that lipid restriction inhibits oocyst sporulation and induces the death of mature oocysts. Therefore, mosquito lipids represent a limiting step in parasite development in the midgut.
Mosquito lipids control Plasmodium virulence to the next mammalian host
As some sporozoites were able to complete their development in spite of massive oocyst die-off (Figure 1D), we examined the effect of lipid starvation on Plasmodium transmission. In these experiments, naïve mice were exposed to the bites of P. berghei-infected control or Lp-deficient mosquitoes. We found that only 20% of mice (n=10) bitten by Lp-depleted mosquitoes became blood-stage positive, as compared to 90% in controls (n=10) (Figure 3A and 3B). To assure that this striking difference in infection levels was not due to the differences in sporozoite loads produced by control and Lp-deficient mosquitoes (Figure 1D), naïve mice were subcutaneously injected with equal numbers of sporozoites isolated from control and Lp-depleted mosquitoes. Again, fewer than 40% of mice (n=18) injected with lipid-deprived sporozoites became infected as compared to 100% in controls (n=16) (Figure 3C). All mice that became infected with sporozoites from Lp-deficient mosquitoes displayed a 1-day delay in the onset of parasitemia (Figure 3D). In addition, only half of the infected mice displayed the severe neurological symptoms typical of experimental cerebral malaria (ECM), and there was a significant delay in ECM development (Figure 3E) compared to control mice, which had 100% incidence of ECM. Overall, mosquito lipid deprivation resulted in a cumulative loss of parasite virulence (Figure 3F). These results strongly suggest that disrupting lipid transport in the mosquito impacts development of Plasmodium liver stages and reduces the initial burden of blood stage parasites, thereby preventing initiation of severe forms of the disease.
Supplemental blood feeding rescues parasite infectivity
To obtain direct evidence of the trans-stadial effect of mosquito lipid deficiency on the Plasmodium liver stages, we gauged development of extra-erythrocytic forms (EEFs) in the hepatoma HepG2 cell line in vitro. Cells were exposed to equal numbers of sporozoites isolated from control and Lp-depleted mosquitoes (Figure 4A) and the development of EEFs was analyzed by fluorescence microscopy two days later. Consistent with the in vivo results, lipid restriction reduced sporozoite infectivity in hepatocytes 4-fold, to levels comparable to those observed after treatment of control sporozoites with cytochalasin D, a potent inhibitor of actin polymerization and Plasmodium invasion (Mota et al., 2001) (Figure 4C). Moreover, significantly smaller EEFs were produced by lipid-deprived sporozoites as compared to controls (Figure 4E and G, Figure S3), suggesting a trans-stadial effect of lipid deficiency on EEF growth. Therefore, we hypothesized that providing lipids directly to the midgut at the critical time of oocyst sporogony might restore sporozoite infectivity and EEF development. To test this hypothesis, P. berghei-infected control and Lp-depleted mosquitoes were blood fed 7 dpi, when the first effects of lipid starvation on oocyst size were observed (Figure 2A). Strikingly, even in the absence of Lp (Figure S4), this supplemental feeding partially restored sporozoite numbers in the mosquitoes (Figure 4B), and fully rescued sporozoite infectivity (Figure 4D) and EEF development (Figure 4F and H). As expected, additional blood feeding did not restore ovary development (data not shown), confirming the importance of Lp in lipid trafficking between mosquito organs. Taken together, these results establish a direct link between lipid starvation at the oocyst stage and Plasmodium infectivity to the next mammalian host.
Mosquito lipids fuel sporozoite mitochondrial membrane potential
We next examined potential mechanisms that attenuate sporozoite infectivity to the next mammalian host. Microscopic inspection of the overall sporozoite morphology by scanning electron microscopy failed to detect any major differences between control and lipid-deficient sporozoites (Figure S5A). Similarly, no major differences were detected in transcript levels of major regulators of sporozoite development (Figure S5B). Since neutral lipids such as cholesterol regulate trafficking of GPI-anchored proteins to the plasma membrane (Zurzolo and Simons, 2016), we gauged surface distribution of the circumsporozoite protein (CSP), the main surface antigen crucial for sporozoite morphogenesis and infectivity (Menard et al., 1997). Immunofluorescence analysis revealed no differences in the distribution of CSP signal at the surface of lipid-deprived sporozoites compared to controls (Figure S5C). In contrast, a striking effect of lipid deprivation was observed on the sporozoite mitochondrial membrane potential as measured by signal intensity of a lipophilic cationic dye (TMRE) whose accumulation in the mitochondrial membrane matrix space correlates with the mitochondrial membrane potential (Perry et al., 2011). We first validated TMRE staining of sporozoites by measuring signal intensity before and after treatment with a mitochondrial uncoupling drug CCCP and showed that drug treatment abolished TMRE signal (Figure S5D). We next assessed TMRE intensity by imaging flow cytometry analysis of sporozoites. Strikingly, a significant decrease in mitochondrial membrane potential was observed in lipid-deprived sporozoites as compared to controls (Figure 5A, B and D). To examine whether reduced mitochondrial activity mitigated sporozoite survival, we gauged levels of transgenic GFP expression as a proxy of parasite live status (Yilmaz et al., 2014). Although GFP expression levels differed between experiments, absence of significant differences in GFP signal between the two groups confirmed that all analyzed sporozoites were alive (Figure 5A, C and E). These results demonstrate that mosquito lipids regulate the mitochondrial activity of sporozoites, and suggest that the decrease in mitochondrial membrane potential is responsible for the attenuated infectivity of the lipid-deprived sporozoites in the mammalian host.
DISCUSSION
Anopheline mosquitoes provide Plasmodium parasites a nutritional niche to complete their life cycle. Here we report the crucial role of mosquito lipids in Plasmodium metabolism that ultimately regulates parasite infectivity and virulence in the mammalian host. We show that parasites hijack the mosquito nutrient transporter Lp for oocyst provision with neutral lipids. These results are in line with previous reports that observed an association of insect Lp particles with Apicomplexan parasites (Atella et al., 2009; Folly et al., 2003; Ximenes Ados et al., 2015). We demonstrate that mosquito lipids regulate multiple stages of Plasmodium development. First morphological anomalies were detected during oocyst sporogony and were mainly characterized by electron-light cytoplasmic vacuolization. Vacuolization is a hallmark of autophagy, a physiological process occurring in normal conditions in a large proportion of Plasmodium liver stages (Eickel et al., 2013). Indeed, under unfavorable circumstances such as starvation, liver stage parasites initiate an autophagy-like cell death to reduce competition and ensure the survival of the fittest progeny (Eickel et al., 2013). As in our experiments, a small proportion of oocysts also displayed similar abnormalities in controls, we propose that autophagy is not restricted to the intracellular liver stages, but is a general mechanism of parasite nutrient sensing, exacerbated in our experiments by lipid restriction. Together with the data accumulated for mammalian stages (Grellier et al., 1991; Itoe et al., 2014; Lauer et al., 2000; Sa et al., 2017), our study uncovers the obligatory Plasmodium dependence on host lipids during throughout its life cycle.
In contrast to sporogonic development, lipid restriction did not induce major morphological or transcriptional changes in the transmissible sporozoite stage. Instead, we revealed an unexpected association between mosquito lipid transport and sporozoite mitochondrial membrane potential. Mitochondria are essential eukaryotic organelles that integrate cellular metabolic signals, generate energy and synthetize intermediates of vital macromolecules. In line with our results, Plasmodium mutants for mitochondrial pathways are defective in oocyst growth and sporulation (Boysen and Matuschewski, 2011; Goodman et al., 2016; Hino et al., 2012; Oppenheim et al., 2014). Moreover, the widely used antimalarial drug atovaquone, which targets the cytochrome bc1 complex and collapses the mitochondrial membrane potential in Plasmodium blood stages (Siregar et al., 2015; Srivastava et al., 1997) also arrests oocyst growth and sporulation, and reduces the number and size of developing EEFs in vitro (Azevedo et al., 2017; Davies et al., 1993; Fowler et al., 1995). Taken together with these results, our findings establish the crucial role of mosquito lipids in parasite mitochondrial function, which shapes Plasmodium sporogonic development and infectivity.
Attenuation of Plasmodium sporozoites is a leading strategy for malaria whole-organism vaccine development (Butler et al., 2012). So far, this has been achieved by irradiation (Seder et al., 2013) or by genetic manipulation (van Schaijk et al., 2014). The data presented here add a new paradigm of virulence attenuation by manipulation of mosquito lipid trafficking. Interestingly, high variability in virulence of P. falciparum sporozoites has been reported under laboratory conditions (March et al., 2013) and in the field (Gupta et al., 1994; Mackinnon and Read, 2004). We propose that disparity in lipid resources between larval breeding sites drive fluctuations in mosquito nutritional status (Gillies, 1954), and thereby may account for the observed variability in Plasmodium virulence in the field. Therefore, the demonstrated direct correlation between sporozoite quantities and transmission efficiency in naïve and vaccinated hosts (Churcher et al., 2017) should also consider sporozoite quality, both of which depend on the mosquito nutritional status. Finally, targeting host/vector lipid trafficking may offer new potent and powerful strategies to block parasite development in the next host, thereby preventing the vicious cycle of malaria transmission.
AUTHOR CONTRIBUTIONS
GC and EAL conceived the study and designed the experiments. GC, ME, RL, CG, and VB performed experiments. AEH and RS contributed reagents and expertise. GC, ME, RL, RS, CG, VB, and EAL analyzed the data. GC and EAL wrote the manuscript.
EXPERIMENTAL PROCEDURES
Mosquito Rearing and Parasite Infections
Anopheles gambiae s.l. mosquitoes were used throughout the study. G3 and 7b lines - an immunocompromised transgenic mosquito line of G3 origin in which expression of the complement-like gene TEP1 is down-regulated (Pompon and Levashina, 2015), were maintained at 29°C 70%–80% humidity 12/12 h day/night cycle. In P. falciparum infections, mosquitoes were fed at 37°C for 15 min through a membrane feeder with NF54 gametocytes cultured with O+ human red blood cells (Haema, Berlin), and thereafter kept in a secured S3 laboratory according to the national regulations (Landesamt für Gesundheit und Soziales, project number 411/08). For P. berghei experiments, mosquitoes were fed on anesthetized CD1 mice infected with the P. berghei GFP-con 259cl2 clone (ANKA strain) that constitutively expresses GFP (Franke-Fayard et al., 2004). Shortly after infections, unfed mosquitoes were removed, while fed mosquitoes were maintained at 26°C for 11-14 days (P. falciparum) or at 20°C for 7-21 days (P. berghei), and then used for the midgut and/or the salivary gland dissections.
RNAi silencing
Double stranded RNA (dsRNA) against lipophorin (dsLp) was produced as previously described (Rono et al., 2010). For RNAi silencing, 1-2 day-old females were anesthetized with CO2 and injected with 69 nl of 3 μg/μl dsLacZ (control) or dsLp using a Nanoject II Injector (Drummond). Mosquitoes recovered for 3-4 days following injection before infections. Efficiencies of RNAi silencing are summarized in Figure S5.
Immunofluorescence Analysis
Mosquito midguts were dissected in PBS, fixed in 4% formaldehyde and washed in PBS. Tissues were permeabilized with Triton X-100 0.1% and incubated with a 1:1 mix of mouse monoclonal anti-ApoI and anti-ApoII antibodies (1/300, clones 2H5 and 2C6) (Rono et al., 2010) overnight at 4°C followed by incubation for 40 min at room temperature with the secondary Cy3-labeled antibodies at 1/1,000 (Molecular Probes), or with 0.1 μg/ml of the Nile Red dye (Sigma Aldrich). Nuclei were stained with DAPI (1.25 μg/ml, Molecular Probes) for 40-60 min at room temperature. Images were acquired using an AxioObserver Z1 fluorescence microscope equipped with an Apotome module (Zeiss). Oocyst sizes were measured by hand-designing a circular ROI of randomly selected oocysts (identified by GFP for P. berghei or by bright field and DAPI staining for P. falciparum) using ZEN 2012 software (Zeiss). Images were then processed by FIJI software (ImageJ 1.47m).
Transmission and Scanning Electron Microscopy
Infected midguts were dissected in PBS, embedded in a drop of low melting agarose (4%) and fixed overnight in 2.5% EM grade glutaraldehyde at 4°C. Midguts were postfixed in 0.5% osmium tetroxide, contrasted with tannic acid and 2% uranyl acetate, dehydrated, and embedded in epoxy resin. After polymerization, sections were cut at 60 nm and contrasted with lead citrate. Specimens were analyzed in a Leo 906E transmission electron microscope at 100KV (Zeiss) using digital camera (Morada).
P. berghei liver-stage development in vitro
The salivary glands and midguts of P. berghei infected mosquitoes were dissected, sporozoites were collected into RPMI medium (Gibco) 3% bovine serum albumin (Sigma Aldrich) and enumerated with a hemocytometer. The P. berghei liver stages were cultured in vitro in HepG2 hepatoma cells and analyzed using standard techniques (Silvie et al., 2008). In short, 15,000 - 20,000 HepG2 cells per well (70% confluence) were plated in the transparent-bottom 96-wells plates (Nalgene International) and incubated for 24 h, then seeded with 10,000 sporozoites and left at 37°C for 2 h. Non-invading sporozoites were removed by washing. As a negative control, sporozoites isolated from control mosquitoes were pre-treated with the inhibitor of actin polymerization Cytochalasin D (Sigma Aldrich) for 10 min. At 48 h post seeding, the cells were fixed for 10 min with 4% formaldehyde, washed with PBS and blocked with 10% fetal calf serum in PBS. Development of the liver-stage parasites was examined using monoclonal mouse anti-GFP antibodies (1/1,000, AbCam) and revealed with the secondary AlexaFluor488 conjugated anti-mouse antibody (1/1,000, Molecular Probes). Nuclei were stained with DAPI (1.25 μg/ml, Molecular Probes). Images were recorded directly on the 96-wells plate using an AxioObserver Z1 fluorescence microscope equipped with an Apotome module (Zeiss) and analyzed for number and size of liver forms using the Axio-Vision ZEN 2012 software (Zeiss).
P. berghei mouse infections in vivo
Mice were housed and handled in accordance with the German Animal Protection Law (§8 Tierschutzgesetz) and both institutional (Max Planck Society) and national regulations (Landesamt für Gesundheit und Soziales, registration number H 0027/12). To determine sporozoite infectivity to mice, the sporozoites collected from the mosquito salivary glands (5,000 sporozoites/mouse) were injected subcutaneously into the tails of 8-10-week-old C57BL/6 females. Bite-back experiments were performed by feeding the P. berghei-infected mosquitoes (18 dpi) on anesthetized naïve C57BL/6 mice (mean 9 bites/mouse, range 2-15). Blood cell parasitemia was determined by daily Giemsa staining of thin blood smears and FACS analysis of the red blood cells with the GFP-expressing blood stage parasites. Infected mice were monitored every 6-12 h for the appearance of severe neurological and behavioral symptoms typical of experimental cerebral malaria (ECM) such as hunched body position, grooming alteration, ataxia, paralysis, or convulsions (Lackner et al., 2006) or by the rapid neurological and behavioral test (RMCBS) (Carroll et al., 2010). All mice with ECM symptoms or the RMCBS score equal or below 5/20 were sacrificed immediately.
Sporozoite imaging flow cytometry
Salivary gland sporozoites were purified in RPMI medium (Gibco) 3% bovine serum albumin (Sigma Aldrich) and kept on ice until staining. Sporozoites were diluted in PBS to concentrations corresponding to 3 mosquito equivalents and stained in the dark with 5 nM TMRE (Cell Signalling Technology) for 20 min at 20°C. CCCP at 50 μM was used as a negative control (see Figure S5D). Sporozoite images were acquired without washing using an ImageStreamX Mk II (Merck Millipore) with a 60x objective over 1 h period. To avoid a possible bias due to the variable pre-acquisition waiting times on ice, the order of sample acquisition was swapped in half of the experimental replicates. GFP-expressing sporozoites were gated by size (Area_M01 or Area_M02) and by GFP intensity (Intensity_MC_Ch02). The analysis was performed with IDEAS 6.2 (Merck Millipore) and FCS files were exported and analyzed by FlowJo v10. Given the small size of the sporozoite mitochondria, the feature finder tool from IDEAS software was used to improve gating of optimally focused sporozoites. The following parameters were used to discriminate focused single sporozoites from debris, and to identify sporozoites sharply focused enough to resolve their mitochondria: H Entropy Std_M01_Ch01_15 over H Variance Std_M01_Ch01, Aspect Ratio_M02 over Symmetry 4-Object(M02,Ch2)_Ch02, Gradient RMS-M01-Ch01 over H Correlation Mean-M02-Ch02_3, and Gradient RMS-M02_Ch02. Briefly, these parameters respectively measure brightfield image texture, GFP image shape, and brightfield and GFP focus. For measurement of mitochondrial membrane potential and GFP intensity, each sample was time-gated to exclude events with prolonged live staining. The parameter corresponding to the lowest background signal in unstained sporozoites was Bright Detail Intensity R7_M04_Ch04, Geo Mean which was selected as bona fide TMRE intensity readout (see Figure S5D). For GFP intensity the Intensity_MC_Ch02, Geo. Mean was used.
Statistical analysis
Statistical analysis were performed using GraphPad Prism 7 software and p values < 0.05 were considered significant (*: p<0.05; **: p<0.001; ***: p<0.0001) and indicated in the figures. The specific tests used are indicated for each figure in the corresponding legend.
ACKNOWLEDGEMENTS
This work was supported by EC FP7 EVIMalaR (grant agreement n°242095) and MALVECBLOK (grant agreement n°223601). The authors thank H. Krüger, M. Andres and L. Spohr for mosquito rearing, mouse work assistance, and Plasmodium infections, and D. Tschierske and D. Eyermann for P. falciparum cultures. Editorial support of Dr. R. Willmott (Bioscript) is gratefully acknowledged. The authors express their gratitude to Prof. K. Matuschewski and his group for continuous fruitful discussions and to Dr. M.M. Mota for helpful comments.