Remodeling of the malaria parasite and host human red cell by vesicle amplification that induces artemisinin resistance

Abstract

Artemisinin resistance threatens worldwide malaria control and elimination. Elevation of phosphatidylinositol-3-phosphate (PI3P) can induce resistance in blood stages of Plasmodium falciparum The parasite unfolded protein response (UPR) has also been implicated as a proteostatic mechanism that may diminish artemisinin-induced toxic proteopathy. How PI3P acts and its connection to the UPR remain unknown, although both are conferred by mutation in P falciparum Kelch13 (K13), the marker of artemisinin resistance. Here we used cryoimmunoelectron microscopy to show that K13 concentrates at PI3P tubules/vesicles of the parasite's endoplasmic reticulum (ER) in infected red cells. K13 colocalizes and copurifies with the major virulence adhesin PfEMP1. The PfEMP1-K13 proteome is comprehensively enriched in multiple proteostasis systems of protein export, quality control, and folding in the ER and cytoplasm and UPR. Synthetic elevation of PI3P that induces resistance in absence of K13 mutation also yields signatures of proteostasis and clinical resistance. These findings imply a key role for PI3P-vesicle amplification as a mechanism of resistance of infected red cells. As validation, the major resistance mutation K13C580Y quantitatively increased PI3P tubules/vesicles, exporting them throughout the parasite and the red cell. Chemical inhibitors and fluorescence microscopy showed that alterations in PfEMP1 export to the red cell and cytoadherence of infected cells to a host endothelial receptor are features of multiple K13 mutants. Together these data suggest that amplified PI3P vesicles disseminate widespread proteostatic capacity that may neutralize artemisinins toxic proteopathy and implicate a role for the host red cell in artemisinin resistance. The mechanistic insights generated will have an impact on malaria drug development.

Figure 1.

Dynamics and localization of PI3P in P falciparum–infected red cells. (A-B) In live P falciparum–infected red cells (3D7 strain; Pf3D7), transgenic expression of a secretory form of the PI3P-binding protein EEA1 fused to mCherry (SS-EEA1^WT-mCherry; red) reveals secretory PI3P in a perinuclear region in early ring parasites (A). Not shown here, but as previously shown, a single-point mutant of EEA1 that fails to bind PI3P is secreted to the parasitophorous vacuole (PV), marked by the dotted circles in the middle panels, where the fluorescence image is merged with the bright field (B). Perinuclear localization is also seen in later schizont stages. The boxed regions in the left-hand panels are magnified in the right-hand panels. (C) Time-lapse images of the parasite’s extracellular merozoite stage (arrowhead) invading red cells to become an intracellular ring (arrow). (D) In live Pf3D7, the transgenic cytosolic form of the PI3P-binding protein EEA1 (cEEA1^WT–green fluorescent protein) is seen associated with punctate vesicles and organelles of late trophozoites/schizonts (as was previously reported) and distinct from perinuclear foci seen for SS-EEA1^WT-mCherry (shown in panels A-B). The boxed region in the left-hand panel is magnified in the right-hand panel. Parasite nucleus stained with Hoechst 33242 (blue). Scale bar, 5 µm. Live cells were imaged using indicator-free RPMI1640 (Gibco) by DeltaVision Deconvolution microscopy with a 100×, NA-1.4 objective on an Olympus IX inverted fluorescence microscope on a temperature-controlled stage at 37°C and a Photometrics cooled custom CCD camera (CH350/LCCD) driven by DeltaVision Software from Applied Precision Inc. (Seattle, WA). (E) Cryo-IEM of PfNF54 (wild-type) parasites probed with antibodies to PI3P and secondary antibody 10-nm gold conjugate. Gold particles are detected in the apicoplast, food vacuole, and tubules (suggestive) of ER. Black arrows indicate PI3P in lumen of tubule; yellow arrows indicate cytoplasmic PI3P; double yellow arrows indicate PI3P vesicular clusters cytoplasmic to ER tubules. Experimental replicates, n = 3. Scale bar, 100 nm. Imaged in a Philips CM120 Electron Microscope (Eindhoven, The Netherlands) under 80 kV. BF, bright field; cEEA1, cytosolic form of the PI3P-binding protein EEA1; GFP, green fluorescent protein; P, parasite nucleus; R, red cell.

Figure 2.

Stage-specific expression and localization of K13 in relation to ER-PI3P, PfBiP, and PfEMP1. (A) Custom antibodies to K13 (supplemental Methods) detect an 83-kDa band in Pf3D7-infected red cells (IRs) but not uninfected red cells (URs) in Western blots (molecular weights in kDa) and localize K13 (green) by IFA in trophozoite and schizont stages (counterstained for host band 3, red), as imaged with a 100×, NA-1.4 objective on an Olympus IX inverted fluorescence microscope using DeltaVision Deconvolution microscopy. (B-C) Cryo-IEM of PfNF54K13WT dually probed for PI3P (6 nm gold) and ER marker BiP (15 nm gold) (B) or K13 (15 nm gold) (C). Black arrow indicates PI3P in lumen of ER tubule; yellow arrows indicate cytoplasmic PI3P; double yellow arrows indicate PI3P vesicular clusters outside of ER tubules on the cytoplasmic face; red arrow indicates low level of PI3P vesicles devoid of K13. Scale bar, 100 nm. (D) Membrane association of K13. Lysates of Pf3D7 were treated as indicated, separated by centrifugation (15 000 g for 30 min) into membrane pellet and soluble supernatant fractions and probed in Western blots for parasite and host (human) markers. Adding 6 M urea (a strong chaotropic agent) for 30 min at 23°C to parasite cell lysates failed to release K13 from the pellet (of parasite cell lysates), although the cytosolic parasite protein PfFKBP was quantitatively detected in the soluble fraction, suggesting that K13 was membrane associated. Sodium carbonate 100 mM, pH 11.5, for 30 min on ice released K13 from the pellet, suggesting that it was peripherally (but not integrally) associated with membranes (and consistently, K13 was also released by 1% Tx-100 for 30 min at room temperature or 1% SDS for 30 min at room temperature). Human band 3 was only released by 1% Triton or 1% SDS, confirming that it was integrally membrane associated. Molecular weight standards (in kDa) are as shown. (E) Stereological analyses of K13-gold particle distribution by cryo-IEM. Vesicles close to the ER appear to bud from ER tubules. Vesicles distal to the ER may be derived from other organellar membranes and cannot be ascribed solely to the ER. (F) IFA single optical sections localizing K13 in segmenter, merozoite, and ring stages. Scale bars are as shown. (G) Anti-ATS antibodies recognize a band >250 kDa in Western blots (as was expected for PfEMP1) in IRs but not URs. Molecular weight standards (in kDa) are as shown. (H) IFA showing single optical sections colocalizing K13 and PfEMP1 (labeled by anti-ATS) in Pf3D7 segmenter and merozoites. Pearson’s correlation coefficients are as indicated. Experimental replicates, n = 3. Scale bars are as indicated. Parasite nucleus (blue) is stained with Hoechst 33242. HuBand3, Human band 3; P, pellet; PC, Pearson’s correlation coefficient; S, supernatant.

Figure 3.

K13 copurified in a PfEMP1-immunoproteome enriched in proteostatic pathways that include the UPR, implicated in clinical artemisinin resistance. (A) Detection of PfEMP1 (arrow, left panel) and total protein content (right panel) captured by anti-ATS antibodies (but not in its absence at −1°) on covalently conjugated protein G beads. Molecular weight standards are in kDa. *Heavy IgG; **Light IgG. (B) Intensity properties of 2 replicate PfEMP1 immunoproteomes. (C) Distribution of proteins in the “503-proteome” comparing their abundance in each proteome expressed as a ratio (abundance ratio [AR] = abundance in Proteome 1/abundance in Proteome 2). X-axis: AR interval range; y-axis: number of proteins. (D) Hypergeometric analyses of 503-proteome and artemisinin-resistant transcriptome associated with clinical strains (for proteostasis enrichment and more detail, see supplemental Tables 2-4; Figure 3E-G). (E) Distribution of normalized intensity (ni; intensity divided by molecular weight) of the 207-proteome (AR = 0.5-1.99). Reactive oxidative stress complex components are BiP and PDI-11 (ni 5.17e8 and 1.2e7, respectively; supplemental Table 5). For detailed annotation and proteostasis enrichment, see supplemental Table 4 and supplemental Figure 3H-I. (F) Model integrating K13-PI3P tubules/vesicles (black-gray spheres based on findings of Figures 1 and 2 and supplemental Figures 1 and 2) and K13-PfEMP1 immunoproteome (green; based on findings in Figure 3; supplemental Figure 3) in proteostatic mechanisms of vesicular export and protein quality control and folding in the ER and cytoplasm (dotted circle). Experimental replicates, n = 2. Other organelles and key are as shown. HT, host targeting sequence; PEXEL, plasmodium export element; PNEP, PEXEL-negative proteins; PPM, parasite plasma membrane; PVM, parasitophorous vacuolar membrane; ROSC, reactive oxidative stress complex.

Figure 4.

Synthetic elevation of PI3P that induces resistance in absence of K13 mutation yields vesicle immunoproteomes enriched in signatures of proteostasis and clinical resistance. (A-B) Detection of PfEMP1 (arrow) captured by anti-ATS antibodies (but not in its absence; −1°) from PfVPS34myc and mass intensity properties of 2 replicate immunoproteomes. (C-D) Detection of PfEMP1 (arrow) captured by anti-ATS antibodies (but not in its absence; −1°) from immunoproteome from PfVPS34AAAmyc (catalytic dead enzyme in which AAA replaces the VPS34-catalytic active-site residues glutamic acid, arginine, and histidine DRH) and mass intensity properties of 2 replicate proteomes. Molecular weight standards are in kDa. (E) Comparative distribution of proteins in proteomes of VPS34myc and VPS34AAAmyc. (F) Hypergeometric analyses showing enrichment of VPS34myc immunoproteome and upregulated (but not downregulated) transcripts of the clinical artemisinin-resistant transcriptome. (G) Protein distribution analyses indicating that 95% of Pf3D7 of proteins enriched in the clinical transcriptome are present in the PfVPS34myc PfEMP1 immunoproteome. (H) Number and percentage of hits for indicated chaperone networks in the vesicle immunoproteome from PfVPSmyc reveals substantial association with reactive oxidative stress complex and T-complex chaperones also known as TCP1 ring complex of the UPR. Mu, mass units.

Figure 5.

K13C580Y, the major mutation of artemisinin resistance, amplifies PI3P tubovesicles, propagating them throughout the parasite and into the red cell. (A-B) Cryo-IEM of artemisinin-sensitive PfNF54K13WT (A) and artemisinin-resistant PfNF54K13C580Y (B) late trophozoite/schizont parasites probed for PI3P (10 nm gold). (C) Region of ER tubules/vesicles. Scale bars, 100 nm (bar sizes differ on basis of magnification). (D) Quantitation of gold particles associated per parasite vacuole for indicated numbers of parasites for each strain. (E-F) PfNF54K13WT and PfNF54K13C580Y ER dually probed for PI3P (6 nm gold) and the secretory ER marker BiP (15 nm gold) (E) or PI3P (6 nm gold) and K13 (15 nm gold) (F). Boxes in left-hand panels are shown magnified in the right-hand panels. Black arrows indicate luminal (space) PI3P; double yellow arrows indicate PI3P vesicular clusters adjacent to K13; red arrowheads indicate small PI3P clusters, and double red arrowheads indicate large PI3P clusters devoid of K13. Scale bars, 150 nm. (G-H) Stereological analyses of percentage of PI3P gold particle density in PfNF54K13WT or PfNF54K13C580Y with each strain normalized to itself (G) and comparative fraction of PI3P-gold particles in each strain (H), accounting for threefold increase in PfNF54K13C580Y in relation to PfNF54K13WT (total number of gold particles in PfNF54K13WT set to 100). (I) RNA sequence analyses for K13C580Y in PfNF54K13WT and PfNF54K13C580Y. Y-axis indicates arbitrary intensity units. (J) Western blots showing levels of K13, PfFKBP protein (a parasite cytosolic protein that serves as a loading control), and associated ratios, indicating no significant difference in levels of K13 protein expression in PfNF54K13C580Y in comparison with PfNF54K13WT. (K-L) PfNF54K13C580Y-infected red cells probed for PI3P (10 nm gold). Red arrows show PI3P at the parasite plasma and vacuolar membranes, in vesicles in the host red cell (scale bar, 500 nm) (K) and associated with “Maurer’s clefts,” intermediates in export to the red cell membrane (scale bar, 100 nm) (L). PI3P is not detected in the host in PfNF54K13WT-infected red cells. Experimental replicates, n = 2. IEM imaged in a Philips CM120 electron microscope (Eindhoven, The Netherlands) under 80 kV. PM, plasma membrane; WT, wild-type.

Figure 6.

Effect of K13 mutation and drug exposure on PfEMP1 expression and export. (A) Schematic of parasite drug exposure. (B) RNA sequence analyses of PfEMP1 expression in 0- to 3-hour PfNF54K13WT and PfNF54K13C580Y rings exposed to 4 nM DHA (positive, red) or mock treated (negative, blue) for 6 hours. Intensity units are as follows: y-axis, PfEMP1 gene id; x-axis, black arrows indicate a major transcript seen in WT parasites. The main PfEMP1 transcript 1200600 (var2csa) was expressed in both wild-type and mutant parasites +/−DHA. Second and third K13WT transcripts (0425800, 0712300) were also expressed in DHA. In C580Y, 0425800, 0400400, 0600200, 0800200, and 0300300 were expressed +/−DHA. PfEMP1 transcript levels were sustained an hour after DHA was washed out (supplemental Figure 5A-B), and parasites successfully matured through subsequent stages of the asexual life cycle (not shown). (C-D) DHA 4 nM potently inhibits PfEMP1 export (green; detected by ATS antibodies in IFA) to the red cell in artemisinin-sensitive PfNF54K13WT but not resistant PfNF54K13C580Y (C). Quantitative analyses of 400 optical sections through 30 infected red cells (D). (E-F) Quantitative analyses of sensitivity of PfEMP1 export (green) to DHA in the artemisinin-sensitive clinical strain ANL-1 but not its resistant clinical counterpart ANL-2 (C580Y). Western blots reveal equivalent PfEMP1 protein levels with or without DHA. For all samples, pixel intensity at 100% exceeded 6.5 × 10⁶ (AU) (panels C-F). Experimental replicates: n = 2 (panels A-B); n = 2 (panels C-F). PfExp2 (Pf-exported protein 2, red), a parasitophorous vacuolar membrane maker, was used to stain the parasite periphery (panels C,E). Scale bars, 5 µm. Imaged with a 100×, NA-1.4 objective on an Olympus IX inverted fluorescence microscope using DeltaVision Deconvolution microscopy software. P, parasite; R red cell.

Figure 7.

Effect of K13 mutation and drug exposure on export of VAR2CSA, cytoadherence, and models. (A) Antibodies to VAR2CSA detect ∼250 kDa in infected red cells but not in uninfected red cells by Western blots (left; see also supplemental Figure 6A). IFA and fluorescence quantitation (right) show inhibition of VAR2CSA (green) export to the red cell by 4nM DHA in artemisinin-sensitive CS2 strain, in relation to mock treatment. Red, Pf exported protein 2 marker of the parasitophorous vacuolar membrane; Hoechst (blue), parasite nucleus; dotted line, red cell periphery. (B) Parental CS2 parasites or transgenic CS2 expressing K13C580Y show similar levels of adherence to CSA; transgenic expression of K13R539T increases adherence slightly (see supplemental Figure 6C for construction of transgenic lines). (C) Potent inhibition of cytoadherence in parental CS2 by DHA was blocked by in trans expression of K13C580Y and K13R539T. Trans expression of K13WT in CS2 reduced cytoadherence, but DHA further decreased adherence (hatched box; see also supplemental Figure 6D; that HA-tagged K13WT and dominant-negative genes are functional was established by Mbengue et al in 2015). Means (±SDs) from 2 experimental replicates are shown (each with triplicate data points; P values as shown) (panels B-C). Imaged with a 100×, NA-1.4 objective on an Olympus IX inverted fluorescence microscope and quantified using DeltaVision Deconvolution microscopy software. (D) Western blots show 4 nM DHA does not block VAR2CSA expression in the CS2 strain (parental) or in transgenic CS2-expressing HA-tagged K13C580Y or K13R539T. Although transgenic K13WT reduces VAR2CSA levels by 50%, DHA does not cause further reduction. Experimental replicates: n = 2 (panels A-D). (E) Models integrate study findings of K13-dependent PI3P tubovesicular action in parasites sensitive and resistant to artemisinins. Key as shown. Green solid circles are apicoplasts; yellow solid circles are food vacuoles; orange solid circles are mitochondria. Scale bars, 5 µm. Exp-2, Pf exported protein 2 marker of the parasitophorous vacuolar membrane; IR, infected red cell; P, parasite; R, red cell; TriC, TCP1 ring complex; UR, uninfected red cell.