PfATG18 links V-ATPase assembly to endocytic membrane dynamics in malaria parasites Yasmin Schmitz, Moumita Sengupta, Carola Schneider, Tanja Ziesmann, Franziska Hellmold, Ute Distler, Rudolph Reimer, Joachim Michael Matz, Vasant Muralidharan Abstract Malaria parasites replicate inside red blood cells, degrading hemoglobin within a specialized digestive vacuole. Efficient hemoglobin processing is essential for parasite survival and influences antimalarial drug susceptibility. The vacuole constantly fuses with incoming hemoglobin-filled vesicles, yet the mechanisms that balance cargo influx with membrane homeostasis remain unclear. Here, using conditional reverse genetics, quantitative live-cell imaging, and 3D electron microscopy, we characterize the autophagy-related protein 18 of Plasmodium falciparum (PfATG18) as a key regulator of vacuolar membrane dynamics. Loss of PfATG18 caused vacuole fragmentation, accumulation of hemoglobin-filled vesicles, and parasite death. These defects were preceded by broad architectural destabilization of the parasite’s V-ATPase, a proton pump controlling organelle acidification and the vacuole’s fusion–fission equilibrium. Direct interference with its membrane sector phenocopied PfATG18 deficiency. We found that PfATG18 does not interact directly with the proton pump but instead associates with a putative V-ATPase assembly factor and with complexes regulating phosphoinositide balance and vesicle trafficking. The breadth of these interactions indicates a multifaceted role at the vacuolar membrane and a regulatory influence on V-ATPase mediated through associated protein machinery. Although a point mutation in PfATG18 has been linked to artemisinin resistance, its complete knockout did not decrease sensitivity. Instead, it rendered ring-stage parasites hypersensitive to dihydroartemisinin, while leaving their susceptibility to other antimalarial drugs unchanged. Together, these findings establish PfATG18 as a central regulator of endocytic membrane homeostasis, essential for V-ATPase function and asexual parasite proliferation in the human blood. Malaria parasites cause disease by multiplying inside human red blood cells, where they consume hemoglobin, the oxygen-carrying protein of the host cell. Hemoglobin is taken up in vesicles and delivered to a specialized digestive vacuole, where it is broken down to support parasite growth and survival. In this study, we characterized the function of PfATG18, a protein associated with the vacuolar membrane. We found that PfATG18 is essential for keeping the digestive vacuole intact and ensuring its efficient fusion with incoming hemoglobin-containing vesicles. When PfATG18 is inactivated, a protein complex required for vacuole acidification and membrane stability called V-ATPase, becomes destabilized and disassembles. Although PfATG18 does not directly interact with the V-ATPase, it associates with proteins involved in V-ATPase assembly, lipid regulation, and vesicle trafficking, indicating a broader role in coordinating processes at the vacuolar membrane. We also found that loss of PfATG18 increases the sensitivity of young blood-stage parasites to antimalarial drug treatment. Together, our findings identify PfATG18 as a key regulator of digestive vacuole function and parasite survival, providing new insight into a critical vulnerability of the malaria parasite. Introduction Parasites of the genus Plasmodium are the causative agents of malaria. Within the human host, they replicate in erythrocytes, residing within a parasitophorous vacuole membrane []. In this intracellular niche, the parasites progress through the ring, trophozoite and schizont stages before lysing the surrounding membranes to release their progeny. During intraerythrocytic development, the parasite catabolizes up to 80% of the erythrocyte cytoplasm, which is internalized at the cytostomes, invaginations of the parasite plasma membrane and the parasitophorous vacuole membrane [–]. Cytostomes generate double-membrane-bound hemoglobin transport compartments (HTCs) that deliver host cell cytosol to an acidic, lysosome-like digestive vacuole (DV) [,]. Within the DV, hemoglobin is proteolytically degraded into small peptides and free heme []. The peptides are exported into the parasite cytosol for further processing [,], while heme is sequestered into bioinert hemozoin crystals, protecting the cell from oxidative damage [,]. Hemoglobin processing and heme biomineralization are essential for parasite survival in the blood, and many antimalarial drugs depend on hemoglobin catabolism for their activity. These include 4-aminoquinolines such as chloroquine, which interfere with hemozoin formation [,], and artemisinin-based compounds, which are activated by heme [–]. Consequently, the kinetics of heme release strongly influence drug efficacy, and mutations that impair endocytosis or proteolysis can confer resistance [–]. During each intraerythrocytic cycle, the DV is formed de novo from small acidified vesicles []. In the most virulent malaria parasite species, Plasmodium falciparum, these vesicles coalesce into a single large organelle that undergoes frequent membrane fusion events to accommodate the constant influx of endocytic cargo [,]. Fusion between the outer HTC membrane and the digestive vacuole membrane (DVM) delivers a single-membrane-bound hemoglobin package into the DV matrix, where it is lysed to release the cargo [,–]. Similar to other eukaryotic systems, endocytic membrane fusion in malaria parasites depends on a variety of trafficking factors, including small GTPases [], membrane tethering complexes [], and likely SNARE proteins []. The endocytic signature lipid phosphatidylinositol 3-phosphate (PI3P) plays a critical role in this process by recruiting trafficking factors to the membranes of both HTCs and the DV [–]. The continuous delivery of HTC membranes to the DV necessitates mechanisms that prevent excessive expansion of the vacuole. In other systems, this is achieved by transferring lipids to a receiving organelle at membrane contact sites [] or by shaping the membrane into tubules or vesicles that are then pinched off with the help of scission proteins []. The latter was shown to depend on phosphatidylinositol 3,5-bisphosphate [PI(3,5)P2], which is produced from PI3P []. The balance between PI3P and PI(3,5)P2 is therefore believed to play a critical role in regulating delivery and recycling of membrane material along the endocytic pathway. The molecular mechanisms that govern membrane equilibrium at the Plasmodium DV are still poorly understood; however, we recently found that the V-ATPase plays a key role in this process []. The V-ATPase is a hetero-multimeric proton pump that acidifies the DV matrix. It consists of an ATP-hydrolyzing V1 sector facing the cytosol and a membrane-embedded Vo sector responsible for proton translocation []. Beyond its canonical function in vacuolar acidification [,], we found that the Vo sector is critical for maintaining DV integrity. Loss of either the Voa or Voc subunits caused dissociation of the V1 sector from the vacuolar membrane and triggered DV fission []. While the V-ATPase appears to limit DV fission, other proteins may promote it in order to maintain vacuolar membrane homeostasis. Indeed, the P. falciparum DV can form autonomous membrane lobes that detach from the parent organelle [,]. A likely contributor to this process is the parasite’s autophagy-related protein 18 (PfATG18), a member of the β-propeller that bind polyphosphoinositides (PROPPIN) family, which has been implicated in endocytic membrane scission in other organisms. In Saccharomyces cerevisiae, ATG18 consists of WD40 repeats forming a seven-bladed β-propeller, with flexible loops extending from the blades [] (). It is required for vacuole fission and exhibits membrane scission activity in vitro, which depends on insertion of its 6CD loop into membranes and on its binding to PI3P and PI(3,5)P2 []. This activity is enhanced by interaction with the retromer complex, a protein coat that forms endosomal transport carriers []. Disrupting retromer engagement of the mammalian ATG18 homolog, WD repeat domain phosphoinositide-interacting protein 1 (WIPI1), leads to accumulation of elongated endosomes, further supporting a role for the PROPPIN family in endocytic membrane fission []. Importantly, this function of ATG18 is independent of its role in autophagy, which involves the recruitment of ATG2, the scramblase ATG9, and other autophagy machinery to facilitate lipid transfer from the endoplasmic reticulum (ER) to the phagophore [–]. Structural homology modelling using AlphaFold3 suggests that the P. falciparum homolog of ATG18 retains the core architecture of other PROPPINs, including the seven-bladed β-propeller (). PfATG18 also features the 6CD loop required for membrane insertion and scission but lacks the 7AB loop, which typically mediates the recruitment of ATG2 during autophagy []. This is consistent with the apparent absence of homologs for ATG2, ATG9 and other core autophagy machinery in Plasmodium, which has cast doubt on the existence of a canonical autophagy pathway in these highly divergent eukaryotes []. The localization and functions of PfATG18 remain uncertain. The protein has been observed at the DVM, with pronounced accumulations at the interface between the DV and its lobes [,], as well as in vesicular compartments and in proximity to the parasite’s non-photosynthetic plastid, the apicoplast [,]. Reverse genetic experiments in P. falciparum and the related apicomplexan parasite Toxoplasma gondii have suggested a role for ATG18 in apicoplast inheritance []. Nevertheless, multiple lines of evidence support a function for PfATG18 at the DVM: (1) it binds efficiently to PI3P in vitro and relocates to the parasite cytoplasm when its phosphoinositide-binding sites are mutated or when PI3P production is pharmacologically inhibited [], (2) it interacts with the DVM-resident transporter multidrug resistance protein 1 (MDR1) in co-immunoprecipitation experiments [], (3) its expression restores vacuolar fission in ATG18-deficient yeast mutants [], and (4) a T38I point mutation in its coding sequence is linked to slower parasite clearance in patients treated with artemisinin-based antimalarials [,], whose activation depends on hemoglobin catabolism in the DV [–,,]. The uncertainty surrounding the involvement of PfATG18 in autophagy, apicoplast inheritance and vacuolar membrane dynamics prompted us to undertake an in-depth functional characterization of this PROPPIN family member. Using conditional reverse genetics combined with quantitative live cell imaging and 3D electron microscopy, we demonstrate that PfATG18 function is critical for DV integrity and endocytic flux. Moreover, we identify a mechanistic link between PfATG18 and the parasite’s V-ATPase, showing that PfATG18 facilitates proper assembly of the proton pump to maintain vacuolar membrane homeostasis. Results PfATG18 localizes to the DVM To study the functions of PfATG18 during blood stage development, we first generated conditional PfATG18 knockout parasites on the genetic background of the dimerizable Cre recombinase (DiCre)-expressing B11 strain of P. falciparum (3D7) []. Using Cas9-mediated genome editing, two loxP sites were introduced into the PfATG18 locus, one within an intron near the start codon and one downstream of the stop codon [] (). These modifications were designed to allow for DiCre-mediated excision of 89% of the PfATG18 coding sequence upon addition of rapamycin (RAP), removing 337 of the protein’s 380 amino acid residues. Two independent conditional knockout lines were generated; in the atg18-mNG:loxP line, PfATG18 was tagged with mNeonGreen (mNG) at the C-terminus, while in the atg18:loxP line, the protein was left untagged. Site-specific integration of the modifying DNA constructs into the parasite genome was confirmed by diagnostic PCR (). To gain insights into the expression and localization of PfATG18, we studied atg18-mNG:loxP parasites by live fluorescence microscopy, which revealed abundant expression of the protein throughout the intraerythrocytic cycle (). In young ring stages, PfATG18-mNG localized to a slender elongated structure. In more mature stages, the protein was consistently detected in a circular pattern around the central hemozoin cluster, suggestive of the DVM (,). We further engineered the atg18-mNG:loxP parasites to express mCherry (mCh)-tagged plasmepsin II (PM2) or chloroquine resistance transporter (CRT), markers of the vacuolar matrix and membrane, respectively. Live microscopy showed that PfATG18-mNG surrounded the PM2-mCh-positive DV matrix, and overlapped extensively with CRT-mCh at the DVM (,; , ). Only in very few instances did we observe in direct contact with the DV an additional PfATG18-mNG positive compartment harboring no CRT-mCh or PM2-mCh (; ), perhaps reflecting DV sub-compartmentalization or dynamic recruitment of PfATG18 to HTCs prior to fusion. However, the low frequency of these observations indicates that PfATG18 occupancy of this compartment, or the compartment itself, is transient. PfATG18 is not enriched at DVM junctions It was previously proposed that PfATG18 is specifically recruited to the membrane junctions between the DV and its lobes []. To test this, we recorded atg18-mNG:loxP + crt-mCh parasites harboring lobed vacuoles. By performing a pixel-by-pixel subtraction of the normalized PfATG18-mNG and CRT-mCh input signals, we visualized the pools unique to each of the proteins. CRT-mCh showed unique weak signals in the parasite cytoplasm and in the vacuolar lumen, most likely corresponding to ER-resident trafficking intermediates and porphyrin-mediated autofluorescence, respectively (). Except for the rare transient structures mentioned above, PfATG18-mNG occupied no areas in the cell that were not also occupied by CRT-mCh. Importantly, the contact sites between the DV and its lobes displayed no specific enrichment of PfATG18-mNG (). We then measured the maximum PfATG18-mNG fluorescence intensity at three different zones of the DVM: at the mother vacuole, at the lobe and at their shared junction (). The signal at the junctional sites was approximately twice that of the individual sub-compartments, as expected for close apposition of two identical membranes (). Accordingly, the signal intensity at the membrane junction was equal to the sum of the fluorescence values of both sub-compartments. Together, these observations indicate no enrichment of PfATG18 at the DV-lobe interface beyond that predicted by close membrane apposition. PfATG18 does not colocalize with endosomal, retromer or apicoplast markers Motivated by previous reports of yeast and mammalian ATG18 homologs functioning in endosomal membrane homeostasis and retromer engagement [], we sought to determine whether a pool of PfATG18 localizes to compartments along the endosomal pathway or to sites of retromer assembly. To that end, we expressed mCh or mScarlet (mSc)-tagged reporter proteins on the genetic background of the atg18-mNG:loxP parasite line for live colocalization analysis. We detected no convincing overlap of PfATG18-mNG with Rab GTPases of early and late endosomes (,) or with the retromer components Vps26 and Vps29 (,), which all displayed punctate or patchy distributions throughout the parasite cytoplasm. In contrast to earlier reports [], we detected no association of PfATG18-mNG with the apicoplast, as indicated by the lack of colocalization with folate transporter 2 (FT2, ) []. In schizont-stage parasites, we frequently observed additional PfATG18-positive punctate compartments. To determine their identity, we colocalized PfATG18-mNG with all previously described markers throughout asexual blood-stage development, including schizonts, but again found no convincing overlap with endosomal small GTPases, retromer components, or FT2 (). In contrast, a faint PM2 signal was detected in the PfATG18-positive puncta, although the vast majority of the protein localized to the DV matrix. This secondary pool of PM2 became apparent only when the display range was adjusted to oversaturate the stronger DV signal (). These compartments may therefore represent trafficking intermediates en route to the DV or storage compartments formed in preparation for the next infection cycle. Collectively, our comprehensive localization analysis indicates that PfATG18 localizes to the DVM and to the membranes of compartments that are either associated with or belong to the DV. PfATG18 is essential for asexual blood stage development To study the consequences of PfATG18 inactivation, we induced DiCre-mediated genomic excision of the PfATG18 coding sequence by supplementing tightly synchronized atg18-mNG:loxP parasite cultures with 20 nM RAP from the ring stage onward. This resulted in efficient excision of the loxP-flanked genomic sequence by the end of the first intraerythrocytic cycle, as shown by diagnostic PCR (). Western blot analysis and live cell imaging indicated that PfATG18-mNG protein was strongly depleted upon RAP treatment, resulting in only ~6% of its physiological expression levels by 36 hours post invasion (hpi) (-). By contrast, expression of CRT-mCh remained completely unaffected by the absence of PfATG18 in the corresponding double mutant (). In flow cytometry-assisted growth assays, dimethyl sulfoxide (DMSO)-treated atg18-mNG:loxP control parasites multiplied efficiently, whereas RAP treatment caused a complete halt in parasite replication (). Off-target effects of the inducing compound can be dismissed, as RAP supplementation did not measurably affect growth of the parental B11 strain (). Microscopic inspection of Giemsa-stained blood smears revealed that PfATG18-deficient parasites arrested their development at young schizont stage, which was accompanied by striking morphological abnormalities including cytoplasmic vacuolarization and a scattered distribution of hemozoin crystals throughout the parasite cell (). We thus conclude that PfATG18 is essential for asexual parasite replication in the blood. DVs lacking PfATG18 undergo fission but do not lyse Using combined polarization and brightfield microscopy of Giemsa-stained blood smears (), we found that the area occupied by the hemozoin crystals remained relatively constant throughout development of DMSO-treated atg18-mNG:loxP control parasites between 24 and 36 hpi (). By comparison, hemozoin crystals in PfATG18-deficient parasites occupied a larger area already at trophozoite stage, and this area increased significantly over time. By 36 hpi, the hemozoin dispersal area of RAP-treated atg18-mNG:loxP parasites exceeded that of DMSO controls by 2.8-fold (). Notably, the drastic differences in subcellular hemozoin distribution did not correlate with a lack of parasite growth, as the area of PfATG18-deficient parasites increased steadily over time, albeit at a slightly reduced rate compared to DMSO controls (). The scattered appearance of hemozoin crystals in in the absence of PfATG18 prompted us to study the morphology of the DV in DMSO and RAP-treated atg18:loxP parasites by live fluorescence microscopy of CRT-mNG. As expected, CRT localized to a single DV in DMSO control parasites. Upon inactivation of PfATG18, however, CRT was associated with a large number of vesicular compartments spread throughout the parasite, indicative of vacuolar fragmentation (). To clarify whether this coincided with a loss of membrane barrier function, we monitored PM2 localization, and found the signal to be restricted to the vacuolar fragments and absent from the parasite cytosol, indicating no breach of DVM integrity upon PfATG18 knockout (). Together, these findings suggest that, unlike its yeast and mammalian counterparts, PfATG18 does not promote vacuole fission, but rather acts to constrain it. Two distinct vesicle populations in PfATG18-deficient parasites We then tested whether the vacuolar fragments of PfATG18-deficient parasites contained host cell cytosol by monitoring the localization of mCh-tagged histidine-rich protein II (HRP2), a soluble parasite protein that is exported into the host cell compartment and accumulates in the DV by endocytic uptake of erythrocyte cytosol []. In DMSO control parasites, HRP2-mCh was found associated with the host cell compartment, but was most strongly concentrated in the DV matrix (). Upon inactivation of PfATG18, the vacuolar HRP2-mCh staining appeared fragmented and we observed several highly fluorescent vesicles that were clearly detached from the mother vacuole. We also detected a second population of vesicles in PfATG18-deficient parasites with much lower HRP2-mCh intensity, comparable to that of the host cell compartment (). To determine whether these vesicles were also derived from the collapsed DV, we visualized internalized host cell cytosol in atg18:loxP parasites expressing CRT-mNG. In this experiment, we chose an alternative strategy for host cell cytosol labelling that exploits the erythrocyte’s capacity to synthesize the fluorescent heme precursor protoporphyrin IX (PPIX) upon supplementation with 5-aminolevulinic acid []. As with HRP2-mCh, PPIX fluorescence was detected in the host cell cytosol and in the DV of DMSO control parasites (). Upon inactivation of PfATG18, the intraparasitic PPIX pool was distributed among numerous vesicular compartments. Not all of these structures were delineated by CRT-mNG, indicating that a considerable fraction of host cell cytosol-containing vesicles formed independently of vacuolar fission (). 3D ultrastructure of PfATG18-deficient parasites The complexity of the PfATG18 knockout phenotype prompted us to study the ultrastructure of DMSO and RAP-treated atg18-mNG:loxP parasites by serial block-face scanning electron microscopy (SBF-SEM) (Supplemental Movies S1 and S2 in ). Three-dimensional reconstruction confirmed that the DV of RAP-treated parasites fragments into numerous hemozoin-containing vesicles without signs of membrane lysis, with an average volume of 0.35 fL (,; and ; Supplemental Movies S3 and S4 in ). In addition, these parasites exhibited a striking accumulation of spherical compartments with a mean volume of 1 fL (). The electron backscatter signal of these compartments was slightly higher than that of the erythrocyte cytoplasm, suggesting they contain host cell hemoglobin (). Consistent with this, previous studies using a membrane-impermeable fluorescent dextran have shown that host cell cytosol is concentrated within vesicular parasite compartments following endocytic uptake []. In very few instances did we observe one or two of the spherical compartments also in DMSO-treated control parasites, indicating that they are physiological but transient. Although the resolution of SBF-SEM does not always allow clear discrimination of closely apposed membranes, we frequently observed inward folds of the spherical compartment envelope, occasionally overlain by a second membrane (). The spherical compartments displayed a range of different membrane features, which we grouped into five categories: (1) slits protruding inward from the periphery, often from a funnel-like invagination; (2) slits terminating in an electron-dense vesicle; (3) peripherally associated electron-dense vesicles without a slit; (4) similarly sized sub-compartments divided by a membrane septum; and (5) electron-lucent inclusions (). Slits and electron-dense vesicles, either alone or in combination, were by far the most common features, while translucent inclusions were rare. All analyzed compartments harbored at least one of these membrane features, with some exhibiting multiple. By contrast, 3D-reconstructed cytostomal invaginations displayed no such modifications, suggesting that they are introduced during or after budding from the cytostome (). Finally, we note that, despite previous suggestions of PfATG18 being involved in apicoplast maintenance, the organelle exhibited comparable size and morphology in both DMSO- and RAP-treated parasites (). Defective HTC fusion blocks hemoglobin processing in PfATG18-deficient parasites We hypothesized that the spherical compartments in PfATG18-deficient parasites represent late hemoglobin transport compartments (HTCs) unable to fuse with the DV, resulting in an inability to process hemoglobin. To test this, we first examined the localization of mSc-tagged Rab7, a late endosomal marker (). In DMSO control parasites, several punctate Rab7-positive compartments were observed. By contrast, PfATG18-deficient parasites harbored an additional population of enlarged Rab7-decorated compartments, often positioned near CRT-positive DV fragments. This pattern was not observed with the early endosomal marker Rab5A (), together indicating that the spherical compartments accumulating in the absence of PfATG18 are matured HTCs that fail to undergo fusion. To confirm the accumulation of unprocessed HTC cargo, protein extracts from saponin-released parasites were analyzed by SDS-PAGE. Coomassie staining revealed that PfATG18-deficient parasites were enriched in a protein of ~15 kDa, which comigrated with the most abundant protein species detected in pure erythrocyte extracts (). Western blot analysis confirmed that this protein was hemoglobin (). As a consequence of defective hemoglobin processing, PfATG18-deficient parasites produced only 62% of physiological hemozoin levels by 36 hpi, as determined by quantitative polarization microscopy (). Together, these findings suggest that while PfATG18-deficient parasites remain capable of endocytosing host cell cytosol, they ultimately fail to deliver the cargo to the DV. DV fragmentation precedes HTC accumulation in PfATG18-deficient parasites We next aimed to dissect the sequence of events underlying the dual phenotype of PfATG18-deficient parasites. To determine whether early DV fragments remain capable of receiving cargo, we treated 28-hour old atg18:loxP + crt-mNG parasites with E64. This broad-spectrum protease inhibitor blocks hemoglobin degradation and induces DV swelling due to accumulation of undigested hemoglobin, unless endocytic trafficking is impaired [,]. Live microscopy of CRT-mNG revealed that 8 hours of E64 treatment caused significant vacuolar bloating in DMSO control parasites (). In RAP-treated parasites, we observed an E64-dependent increase in the size of individual DV fragments. Because the DV of these parasites was already fragmented at the time of E64 addition, these results indicate that early DV fragments had received endocytic cargo during the 8-hour treatment window (). This was further supported by microscopic analysis of Giemsa-stained blood smears, which revealed host cell cytosol accumulated in bloated DV fragments of E64-treated PfATG18-null parasites (,). To allow for the simultaneous monitoring of DV morphology and endocytic hemoglobin delivery, we introduced into the atg18:loxP + crt-mNG parasites an expression cassette for P40PX-mSc, a PI3P-binding fluorescent reporter that marks both the DVM and HTCs [,]. We imaged these transgenic parasites by live fluorescence microscopy at 28 and 36 hpi. At both timepoints, DMSO-treated control parasites exhibited an intact DV, with P40PX-mSc localizing almost entirely to the vacuolar membrane, indicative of efficient PI3P transfer from HTCs to the DV (). Although RAP-treated parasites exhibited a fragmented DV already at 28 hpi, P40PX-mSc was found associated exclusively with CRT at that time. This changed dramatically 8 hours later, when a substantial fraction of P40PX-mSc localized to vesicular compartments devoid of CRT, identifying them as HTCs (). Together, these findings suggest that in PfATG18-deficient parasites, vacuolar fragmentation precedes HTC accumulation. Loss of PfATG18 hypersensitizes ring-stage parasites to dihydroartemisinin Perturbations in hemoglobin handling and DV physiology can both confer resistance to drugs whose activity depends on hemoglobin consumption, such as chloroquine and artemisinin [–]. Reports linking a T38I mutation in PfATG18 to prolonged parasite clearance half-life in artemisinin-treated patients [,] therefore motivated us to examine whether PfATG18-deficient parasites might similarly exhibit reduced drug sensitivity. To test this, we first established conditions that allowed us to quantify parasite replication as a readout in drug sensitivity assays. We delayed RAP induction so parasites could produce minimal amounts of PfATG18 and complete one replication cycle. When RAP was applied at 16 hpi, parasites progressed into the second intraerythrocytic cycle at ~37% efficiency, compared to DMSO controls (). At this induction time, PfATG18 protein levels were reduced by ~90% at the end of the first cycle (). Live-cell imaging revealed a severely fragmented DV without significant accumulation of HTCs, indicating that PfATG18-dependent functions were disrupted in schizonts, although the delayed induction bypassed the fusion defect during the first cycle (). Using this setup, we assessed drug sensitivity by exposing parasites at the time of induction to increasing concentrations of dihydroartemisinin (DHA), chloroquine, or the negative control compound WR99210, which inhibits folate biosynthesis (). Parasite DNA was quantified 68 hpi using SYBR Gold []. The resulting growth-response curves showed no RAP-dependent shifts in IC50 values for any of the drugs (). To address the possibility that drug action might have occurred before loss of PfATG18 function, we repeated the assays with drug addition late in schizogony (). Again, no differences in drug-sensitivity were observed (). To examine whether the loss of HTC-DV fusion (bypassed under the previous induction schedule) could confer drug resistance, we treated cultures with DMSO or RAP immediately after invasion and added DHA 28 hours later (). At that time, the DV was already fragmented and HTC accumulation was imminent (). Quantification of replicated DNA at the end of the first cycle yielded robust drug-response curves, yet again showing no differences in DHA sensitivity (). Because artemisinin resistance is associated with increased ring-stage survival [,], we next induced PfATG18 loss at 16 hpi and treated second-cycle rings with DHA for 6 hours, then assessed their survival (). In DHA-free controls, DMSO-treated parasites matured normally into healthy second-cycle trophozoites, whereas RAP-treated parasites arrested as late rings or dysmorphic young trophozoites, with some parasites appearing dead, indicating that delayed PfATG18 loss already impairs second-cycle development (). A 100 nM DHA pulse caused near-complete death in both conditions, producing uniformly pyknotic forms that were readily distinguishable from surviving parasites (,). A 12 nM DHA pulse reduced survival from 87% to 51% in DMSO-treated parasites and from 64% to 16% in RAP-treated parasites (). After accounting for baseline survival differences, PfATG18-deficient parasites were approximately twice as sensitive to DHA at this stage (). To determine whether this phenotype simply reflected the deteriorating physiological state of PfATG18-deficient parasites, we repeated the ring-stage survival assays using chloroquine and proguanil pulses at concentrations titrated to kill approximately 50% of the parasite population. Intriguingly, the absence of PfATG18 had no effect on survival under these conditions (-). Together, these data demonstrate that PfATG18-deficient rings display a heightened vulnerability to DHA that is not observed with other antimalarial treatments. Loss of PfATG18 results in V-ATPase disassembly We recently reported that DVM integrity depends on the parasite’s V-ATPase complex [] (). Genetic inactivation of individual Vo sector subunits caused disassembly of the proton pump and triggered vacuolar fission – an effect similar to that observed in PfATG18-deficient parasites []. This raises the possibility that PfATG18 may play a role in regulating V-ATPase stability and function. To investigate this, we first examined the localization of mNG-tagged subunit V1B in atg18:loxP parasites. In DMSO-treated controls, V1B localized predominantly to the DVM surrounding the hemozoin crystals. Upon RAP treatment, however, V1B no longer associated with the hemozoin clusters and instead localized diffusely throughout the cell, indicating dissociation from the DVM (). A similar pattern was observed for subunit V1C, suggesting that PfATG18 loss broadly disrupts membrane association across the entire V1 sector (). Because fragmentation hindered reliable identification of the DVM, we further modified the atg18:loxP + v1B-mNG line to express mCh-tagged CRT. Following PfATG18 inactivation, V1B no longer colocalized with CRT at the fragmented DVM (), and quantification of membrane-associated versus cytosolic signals revealed that these V1B pools had fully equilibrated in PfATG18-deficient parasites (). The dissociation of V1 sector subunits indicates defective V-ATPase assembly and has been shown to occur in response to perturbations of either the Vo or the V1 sector []. Importantly, we have shown that DV fragmentation is exclusively associated with Vo sector dysfunction, whereas perturbations of the V1 sector alone do not produce this effect []. We therefore investigated the localization of subunit Voa in PfATG18-deficient parasites and found it to be severely depleted from the vacuolar membrane (). As with V1B, we introduced CRT-mCh as a DVM marker and confirmed that Voa association with CRT-defined DV fragments was markedly reduced, although dissociation was not complete (,). Our findings suggest that V-ATPase disassembly may be the proximal cause of DV fragmentation in PfATG18-deficient parasites. To determine whether disassembly precedes fragmentation, we treated the atg18:loxP + v1B-mNG + crt-mCh line with RAP at 28 hpi and analyzed V1B localization eight hours later, at the onset of DVM fission. At this stage, approximately one half of the parasite population retained an intact DV, while the other half had initiated vacuolar fragmentation. In marked contrast to DMSO controls, V1B was almost completely dissociated from the DVM in RAP-treated parasites, regardless of whether the DV was already fragmented or not, indicating that PfATG18 inactivation disrupts V-ATPase assembly before vacuolar integrity is compromised (, Fig ). To determine whether V-ATPase dysfunction could explain both vacuolar fragmentation and the fusion defect in PfATG18-deficient parasites, we tested whether directly compromising Vo sector integrity recapitulates this dual phenotype. We modified a previously generated conditional knockout line for subunit Voc expressing CRT-mNG [] to additionally express the exported host cell cytosol reporter GBP1301–180-mSc. As previously shown, Voc inactivation caused fission of the DV into numerous CRT-positive compartments [], which contained high levels of GBP1301–180-mSc (). Importantly, the host cytosolic marker was also detected in accumulating spherical compartments lacking CRT, at concentrations comparable to the erythrocyte cytosol, suggesting that they represent hemoglobin trafficking intermediates upstream of the DV (). These findings indicate that inactivation of Vo sector subunits affects both vacuolar integrity and fusion. Lower abundance of fully assembled Vo sectors at the DVM could thus sufficiently explain the dual defects observed in PfATG18-deficient parasites. PfATG18 associates with a putative V-ATPase assembly factor, phosphoinositide-converting enzymes and vesicle trafficking machinery We next sought to determine whether PfATG18 influences V-ATPase function through direct physical interactions or indirectly. To identify potential PfATG18 interaction partners, we immunoprecipitated the protein from atg18-mNG:loxP parasites and analyzed co-purifying proteins by mass spectrometry (see Source Data). After ranking hits by enrichment and statistical significance relative to parental controls, we found that none of the V-ATPase subunits ranked highly (), suggesting that the contribution of PfATG18 to V-ATPase assembly is unlikely to involve direct contact with the proton pump. Instead, among the highest-ranking hits we detected a conserved parasite protein of 721 kDa (PF3D7_0804500) that bears distant resemblance to components of the RAVE/Rabconnectin-3 complexes, which in other eukaryotes function as regulatory scaffolds that mediate reversible Vo-V1 coupling []. We therefore named this protein Rabconnectin-3-like (Rbcn3L). Other top-ranking proteins included two components of the class III phosphoinositide 3-kinase (PI3K) complex (), which synthesizes the endocytic signature lipid PI3P and whose disruption impairs HTC-DVM fusion in P. falciparum [,]. In addition, two subunits of the Fab1/PIKfyve complex were identified, which synthesizes PI(3,5)P2 from PI3P in other eukaryotes [] (). Additional hits included multiple membrane trafficking factors, including a Sec1/Munc18-like protein and several small GTPase regulators (,). Four components of the HOPS/CORVET vesicle-tethering machinery, the phosphoinositide-binding protein PX1, and Myosin F were also enriched – all previously implicated in hemoglobin trafficking [,,]. By contrast, components of the retromer complex, which have been reported to interact with ATG18 in yeast and mammalian cells, ranked low in our analysis, consistent with the observed lack of colocalization (,; ,). Although not all enriched proteins may represent direct PfATG18 interactors, our findings suggest that PfATG18 might function as a versatile interaction hub at the DVM. It may recruit, stabilize, or regulate diverse proteins and assemblies that support core DV functions, including membrane fusion, lipid conversion, and V-ATPase assembly, thereby exerting broad effects on organelle physiology. Discussion Malaria parasites possess only a limited subset of autophagy-related genes and lack key components required for autophagy initiation and autophagosome membrane expansion [,]. Beyond PfATG18 and PI3K, they retain an ortholog of ATG8, along with the machinery necessary for its processing and membrane conjugation. In classical autophagy, ATG8 functions as a central hub, orchestrating autophagosome formation, cargo recognition, and membrane dynamics []. In P. falciparum, however, PfATG8 localizes specifically to the apicoplast, and its absence results in lethal defects in organelle inheritance [,]. Remarkably, survival of PfATG8-deficient parasites can be fully restored by supplying the apicoplast-derived metabolite isopentenyl pyrophosphate, thus bypassing the organelle’s essential metabolic functions []. These findings, alongside a growing body of evidence, indicate that malaria parasites have lost many components of the autophagic machinery during evolution, while others have been repurposed for roles beyond classical autophagy. Similarly, non-canonical functions have been proposed for ATG18 in apicomplexan parasites, but also in other model organisms, including roles in apicoplast inheritance [], retromer engagement [,], and endocytic membrane fission [,]. In our study, however, neither localization, phenotype, nor protein interaction data provided evidence for an involvement of PfATG18 in apicoplast- or retromer-related processes. Instead, we found that loss of PfATG18 leads to multiple cellular abnormalities, all linked to the parasite’s endocytic pathway (). The most striking of these were fragmentation of the DV and accumulation of HTCs, together indicating that PfATG18 directs endocytic membranes toward fusion rather than fission. Accordingly, PfATG18 may facilitate both heterotypic fusion between the DV and HTCs and homotypic fusion between the DV and its lobes. Our findings contrast with reports from yeast and mammalian cells, where ATG18 homologs function in membrane scission: disruption of these proteins prevents vacuoles or endosomes from undergoing fission, causing organelle enlargement [,]. Although DV fragmentation is the exact opposite of these phenotypes, PfATG18 expression was shown to rescue the vacuolar fission defect in ATG18-deficient yeast, highlighting how conserved molecular functions can produce different outcomes depending on cellular context []. In PfATG18-null parasites, HTCs accumulate only after DV fragmentation has already commenced, suggesting that their failure to fuse could be a secondary consequence of a disrupted DV. However, this interpretation is challenged by two observations: (1) early DV fragments can still receive cargo, and (2) other Plasmodium species naturally maintain multiple small DVs yet process hemoglobin just as efficiently [,]. An alternative possibility is that PfATG18 regulates DV maintenance and fusion through distinct pathways by serving as a scaffold for multiple molecular assemblies at the DVM, a functional versatility commonly observed in WD40 repeat proteins []. This notion is consistent with the identification of functionally distinct protein assemblies as potential PfATG18 interactors, including vesicle tethering complexes, phosphoinositide-converting assemblies, and other protein machinery involved in endocytic trafficking. While our Co-IP data identify strong candidates for PfATG18 binding, complementary proximity labeling studies may help determine whether these interactions also occur in the context of the intact parasite cell. Our observation that the V-ATPase disassembles in response to PfATG18 inactivation raises the possibility that the defects in endocytic membrane dynamics may arise directly from proton pump dysfunction. The V-ATPase has been implicated in membrane fusion and fission across different organisms, although it often remains unclear whether these effects are structural or pH-dependent [,]. In yeast, vacuole fusion is facilitated by interactions between opposing Vo sectors in trans, promoting lipid mixing and hemifusion, which likely involves Vo-SNARE interactions [,]. Salt-induced vacuole fission has also been reported to depend on V-ATPase activity []. We show that in P. falciparum, targeted interference with Vo sector integrity phenocopies the dual defects observed in PfATG18 knockout parasites and that disruption of PfATG18 is accompanied by Vo sector destabilization, leading to complete loss of Vo-V1 coupling. Our findings may also provide a mechanistic explanation for previous observations in yeast, where ATG18 was reported to associate cyclically with the vacuole in strict synchrony with organellar pH oscillations [], and to ensure vacuolar pH asymmetry during budding []. We found no evidence for direct interactions between PfATG18 and the V-ATPase, although weak or transient interactions in cellulo cannot be excluded. Instead, one of the most prominent proteins identified in the PfATG18 pull-downs was Rbcn3L, which exhibits distant resemblance to RAVE/Rabconnectin-3 complexes. In other eukaryotes, these proteins facilitate Vo-V1 association, particularly when the two sectors have dissociated in response to cellular stress signals []. While V1 dissociation in PfATG18-null parasites may result from loss of Rbcn3L function, the destabilization of the Vo sector is more difficult to reconcile with this mechanism, unless Plasmodium Rbcn3L has evolved additional roles in V-ATPase assembly beyond sector coupling. It thus appears likely that PfATG18 may influence the proton pump’s membrane sector indirectly by modulating conditions at the DVM that support its stability and function. One such condition could be the homeostasis of phosphoinositide lipids. Indeed, multiple components of the PI3K and Fab1/PIKfyve complexes ranked among the top hits in PfATG18 pull-downs. In yeast, PI(3,5)P2 produced by Fab1 directly binds to the N-terminal domain of Voa which helps anchor the subunit in the membrane and stabilizes Vo-V1 interactions [,]. Similarly, disruption of PI3K-mediated PI3P synthesis, resulted in significantly reduced levels of both Vo and V1 sector subunits in yeast vacuolar vesicle preparations []. It is therefore tempting to speculate that the phenotypes observed in our mutant may largely result from disturbed phosphoinositide homeostasis, with PfATG18 regulating their balance at the DVM by modulating PI3K and Fab1/PIKfyve activity. In yeast, ATG18 has been proposed to act as a negative regulator of PI(3,5)P2 levels. In this model, ATG18 is recruited from the cytosol to the vacuole membrane when PI(3,5)P2 levels spike, often in response to osmotic stress. Elevated PI(3,5)P2 would then promote ATG18 binding to the membrane, where it could downregulate Fab1/PIKfyve activity and limit further PI(3,5)P2 production []. In contrast to yeast, PfATG18 does not appear to exist in a cytosolic pool, arguing against a recruitment-based regulation mechanism and rather supporting a constitutive role at the DVM. While PI3P is highly enriched at the DVM, PI(3,5)P2 has not yet been detected in the parasite, either due to technical limitations arising from its low abundance, or because it may be absent entirely []. In support of the latter scenario, PfATG18 interacted only with PI3P in liposome binding assays, whereas its T. gondii ortholog bound both PI3P and PI(3,5)P2 [], although broader phosphoinositide interactions were observed in protein-lipid overlay assays []. Remarkably, when expressed in ATG18-deficient yeast, PfATG18 reestablishes vacuolar fission, a process previously shown to depend on PI(3,5)P2 [,]. Thus, even if the Plasmodium Fab1/PIKfyve complex has lost its catalytic activity during evolution, its interaction with ATG18 may still be conserved and mediate functions beyond phosphoinositide balance. Mechanistic studies on the parasite’s Fab1/PIKfyve complex, combined with sensitive phosphoinositide biosensors, are thus needed to clarify the role of PI(3,5)P2 in parasite development and its relationship with PfATG18. Our SBF-SEM analysis allowed three-dimensional reconstruction of accumulating HTCs, revealing distinct membrane profiles closely associated with the HTC envelope. These slit-like invaginations frequently terminated in a small intralumenal vesicles whose electron density suggests they contain host cell cytosol. It remains unclear whether these features represent fission scars from HTC budding at the cytostome, or form through membrane inward bending afterwards. The latter scenario would be consistent with our observation that the slits often originate from funnel-like pits, reminiscent of ESCRT-mediated intraluminal vesicle formation []. Vesicular HTC subcompartments have been observed previously by conventional transmission electron microscopy independently of PfATG18 and therefore appear to represent a general feature of HTC physiology [,,,]. Our SBF-SEM dataset provides the first three-dimensional characterization of their ultrastructure; however, higher-resolution approaches such as cryogenic electron tomography will be required to further resolve their organization and membrane architecture. The role of PfATG18 in hemoglobin trafficking offered a plausible mechanistic basis for the association between the T38I point mutation and prolonged parasite clearance in artemisinin-treated patients [,]. We hypothesized that reduced hemoglobin delivery, and consequently diminished heme release in the DV, would decrease artemisinin sensitivity in PfATG18-deficient parasites due to reduced drug activation [,]. However, despite multiple assay variations, including optimized induction and drug incubation schedules, we were unable to support this hypothesis. This is consistent with a previous study examining the isolated effect of the T38I mutation, which reported no changes in sensitivity to artemisinin-based antimalarials, using both standard growth-inhibition and ring-stage survival assays []. It therefore remains possible that resistance effects conferred by the T38I mutation are too subtle to resolve under laboratory conditions but may nonetheless provide a selective advantage in the field. Alternatively, the T38I mutation may confer a selective advantage only in specific genetic backgrounds that harbor additional mutations linked to artemisinin resistance. To our surprise, loss of PfATG18 rendered ring-stage parasites hypersensitive to DHA, but not to other antimalarial drugs. While the mechanistic basis of this observation remains unclear, we hypothesize that this phenotype may reflect disturbed phosphoinositide homeostasis, as elevated levels of PI3P have previously been linked to artemisinin resistance [,]. If PfATG18-deficient parasites exhibit a phosphoinositide balance shifted toward lower PI3P and higher PI(3,5)P2, like their yeast counterparts [], this could explain the observed sensitivity profile. However, a mechanistic link between PfATG18, phosphoinositide homeostasis, and artemisinin resistance remains to be established in future studies. Methods Ethics statement This study does not involve live participants. Blood donations were obtained from the University Medical Center Hamburg-Eppendorf and were fully anonymized. Structural homology modelling An AlphaFold3 model of PfATG18 was aligned to the crystal structure of the ScATG18 β-propeller core []. Structural alignment was performed in PyMol (version 2.5.7, Schrödinger, LLC) using the “cealign” command, and the RMSD value was calculated for the aligned residues. Due to low-confidence predictions, the 4CD and 6CD loops extending from the PfATG18 β-propeller core were excluded from the alignment. Parasite cultivation Plasmodium falciparum parasites were maintained at 37°C in human erythrocytes (B+) suspended in Roswell Park Memorial Institute medium 1640 (RPMI 1640, Thermo Fisher Scientific, 31800089) supplemented with 0.5% AlbuMax II (Thermo Fisher Scientific, 10624053). Cultures were kept in an atmosphere of 1% O2, 5% CO2, and 94% N2, and synchronized as previously described []. Briefly, mature schizonts were isolated by centrifugation over a 60% Percoll cushion (Cytiva, 17089101) and allowed to rupture and invade fresh erythrocytes for 2 hours under constant agitation. Residual schizonts were subsequently removed using a second Percoll gradient, followed by 5% sorbitol treatment (Carl Roth, 6213.1) of the infected cells, yielding highly synchronous ring-stage cultures. Parasitemia was determined either by microscopy of Giemsa (Merck, 1.09204)-stained blood smears or by flow cytometry of SBYR Green (Merck, S9430)-stained cultures and adjusted by dilution with uninfected erythrocytes. Conditional PfATG18 knockout lines were treated with DMSO (Carl Roth, 7029.1) or RAP (Clontech Laboratories, Cay13346) from the ring stage onward, followed by phenotypic analysis at 36 hpi, unless otherwise specified. The voc:loxP + crt-mNG + gbp1301–180-mSc parasites were induced and analyzed as described previously []. E64 (Merck, E3132) and 5-aminolevulinic acid (Merck, A3785) were added to the culture medium at 28 hpi at final concentrations of 21.7 μM or 200 μM, respectively. Generation and validation of transgenic parasites The atg18:loxP parasite line was generated via Cas9-guided homologous integration of a synthetic repair template containing recodonized PfATG18, flanked by two loxP sites – one within an artificial intron and the other immediately downstream of the stop codon [] (). In the atg18-mNG:loxP line, the C-terminus was additionally tagged with mNG. These genetic modifications were introduced into the 3D7-based B11 line, which expresses dimerizable Cre recombinase (DiCre) and permits RAP-induced excision of loxP-flanked genomic sequences []. For Cas9-mediated genomic integration, ~ 108 Percoll-enriched schizonts were electroporated using a Lonza Amaxa 4D Nucleofector device, introducing 20 μg of guide plasmid and 60 μg of linearized repair template. Following transfection, parasites were allowed to invade fresh erythrocytes, and drug selection with 2.5 nM WR99210 (Jacobus Pharmaceutical) was applied 24 hours later for a total duration of 4 days. Transgenic parasite clones were isolated 1–2 weeks later using a previously described plaque assay []. Correct genomic integration of the repair templates and RAP-induced excision of loxP-flanked sequences were verified by diagnostic PCR using primer combinations shown in and . Endogenous tagging of CRT, PM2, HRP2, FT2, Vps26, Vps29, V1B, V1C, and Voa was also achieved by Cas9-mediated genome editing. Clonality of the resulting mutants was confirmed by live fluorescence microscopy. mSc-Rab5a [], mSc-Rab7 [], P40PX-mSc [], and GBP1301–180-mSc [] fusion proteins were expressed episomally under constant drug pressure with 0.9 μM DSM1 (BEI Resources, MRA-1161). Repair templates and Cas9 guide sequences are provided in and , respectively. All transgenic parasite lines used in this study are listed in . Analysis of parasite replication and drug sensitivity To analyze parasite replication, ring-stage cultures were adjusted to an initial parasitemia of 0.1%, and parasitemia was monitored over four intraerythrocytic cycles at two-day intervals. The percentage of infected erythrocytes was determined by flow cytometry of cultures stained for 40 minutes at 37°C in medium containing 1:10,000 SYBR Green, followed by 40 minutes of de-staining in plain medium. Samples were analyzed on an automated Agilent NovoCyte 3000VYB flow cytometer equipped with a NovoSampler Pro. To assess parasite drug sensitivity, synchronized atg18:loxP cultures were adjusted to 2% parasitemia at 2% hematocrit and incubated in 96-well black-bottom plates (Thermo Fisher Scientific, 137101) with DMSO/RAP and varying concentrations of WR99210, chloroquine (Cayman Chemical, 30708), or DHA (AdipoGen Life Sciences, AG-CN2–0468) in a final volume of 200 μL. Cells were lysed at 68 hpi, and DNA was stained with SYBR Gold (Thermo Fisher Scientific, S11494), as described previously []. Fluorescence was quantified using a Tecan GENios plate reader, with excitation and emission wavelengths set to 485 nm and 535 nm, respectively. Raw fluorescence values were corrected for non-infected erythrocyte controls and normalized to the highest and lowest readings. IC50 values were calculated using non-linear regression in GraphPad Prism (version 10.4.1), applying the “log(inhibitor) vs normalized response – Variable slope” equation. To assess ring-stage survival following treatment with DHA, chloroquine, or proguanil, atg18:loxP parasites were induced at 16 hpi. First-cycle schizonts were isolated and added to fresh erythrocytes to generate tightly synchronized second-cycle ring stages at ~5% parasitemia and 2% hematocrit, seeded into 24-well plates at a total volume of 600 μL per well. Ring-stage parasites were then exposed to varying drug concentrations for 6 hours at 37°C. Following the drug pulse, cultures were washed three times with medium, resuspended in 600 μL per well, and incubated at 37°C. Parasite survival was assessed 16 hours later by microscopic examination of Giemsa-stained blood smears, with viable and pyknotic parasites quantified by manual counting. The atg18:loxP mutant was generated in a 3D7-based genetic background lacking the major artemisinin resistance-associated mutations. Therefore, lower DHA concentrations than those typically used in ring-stage survival assays were used (12 nM instead of 700 nM), providing sufficient dynamic range to detect differences in DHA susceptibility. The recovery period following the drug pulse was limited to 16 hours to distinguish DHA-induced parasite death from effects arising from PfATG18 deficiency itself. Light and fluorescence microscopy Light and fluorescence microscopy were performed using a DM6 B microscope equipped with a K8 B/W camera and a K3C color camera, all from Leica. The microscope was fitted with two crossed polarizers, enabling in situ visualization and quantification of birefringent hemozoin crystals. All images were analyzed and processed using FIJI (version 2.16.0/1.54p). Transmitted polarized light and fluorescence signals were measured as mean intensity, maximum intensity or raw integrated density, with appropriate background corrections applied in all cases. All fluorescence imaging was performed with live parasites. To visualize dimmer pools of PPIX, HRP2-mCh, or GBP1301–180-mSc, selected micrographs were intentionally oversaturated. Signals specific to CRT-mCh or ATG18-mNG in the respective double mutant were identified by first normalizing the input signals to their respective minimum and maximum pixel values. The normalized intensities were then subtracted on a pixel-by-pixel basis, and the resulting differences were visualized as pseudo-images. To quantify the membrane association of V1B-mNG and Voa-mNG, fluorescence intensity was measured in randomly selected regions of the DVM and parasite cytoplasm, identified by the presence or absence of CRT-mCh labelling, respectively. Ratios of the maximum intensity values were calculated. Parasite size was determined microscopically in Giemsa-stained blood smear as the percentage of the erythrocyte area occupied. To quantify the hemozoin dispersal area, the polygon selection tool in FIJI was used to manually outline a single region per parasite that enclosed all hemozoin particles, minimizing the number of nodes and avoiding acute polygon angles. The resulting dispersal area was then normalized to parasite size. Parasites were classified as exhibiting E64-induced bloating of the DV or its fragments when hemozoin crystals or crystal clusters were clearly enclosed within a hemoglobin-containing lumen whose coloration was similar to that of the host cell cytosol. Serial block-face scanning electron microscopy For SBF-SEM analysis, infected erythrocytes were enriched 36 hpi via Percoll gradient centrifugation. Cells were fixed in 2% paraformaldehyde (Science Services, E15713) and 2.5% glutaraldehyde (Plano, R1020) in Dulbecco’s PBS (Merck, D8537), then embedded in 6% gelatin (Merck, 1040700500). Post-fixation was performed with 2.5% glutaraldehyde in PBS, followed by staining with 2% OsO4 (Carl Roth, 7436.1), 1.5% potassium ferrocyanide (Carl Roth, 7974.1) and 2 mM CaCl2 (Carl Roth, A119.1) in PBS on ice. Samples were then sequentially incubated with 0.5% thiocarbohydrazide (Merck, 223220), 2% OsO4 and 0.1% gallic acid (Merck, G7384). Subsequent overnight staining was performed at 4°C with 1% uranyl acetate (Merck, 8473). The following day, samples were stained with Walton’s lead aspartate (Carl Roth, HN32.1, 1690.2) at 60°C and dehydrated through a graded ethanol series (Carl Roth, 9065.3) on ice. Samples were infiltrated with 50% and 70% Epon (Carl Roth, 8619.2, 8623.2, 8639.2, 8621.1) in ethanol, followed by two incubations in 100% Epon and polymerization with 3% silver flakes (Merck, 327085) at 60°C. Sample blocks measuring 0.5 x 0.5 mm were cut, mounted, and placed into a Gatan 3View stage within a Jeol JSM-7100F scanning electron microscope. Imaging was performed at 3 kV acceleration voltage with a stage bias of +500 V, using a probe current setting of 1, 3 nm pixel size, and 50 nm section thickness. Acquisition and stack alignment were carried out with Gatan Digital Micrograph (version 3.32). Images were processed in FIJI using the StackReg plugin for alignment and the DenoisEM plugin [] with Tikhonov algorithm for denoising. Segmentation and 3D reconstruction were performed using Microscopy Image Browser [] (version 2.9.1) and Imaris (version 9.8.2, Oxford Instruments). Western blotting For Western blot analysis, parasites were released from host erythrocytes by treatment with 0.03% saponin (Merck, 47036) in PBS for 10 minutes on ice. The parasites were washed several times with PBS until the supernatant was colorless, and then lysed in PBS containing 4% SDS (Carl Roth, 0183.1), 0.5% Triton X-100 (Merck, 108603), and a protease inhibitor cocktail (Merck, 11697498001). Lysates were mixed with Laemmli buffer and incubated at 95°C for 5 minutes prior to SDS-PAGE. Separated proteins were either stained in-gel using Coomassie Brilliant Blue R250 (Merck, 1.12553) or transferred onto nitrocellulose membranes (Cytiva, 10600007). Primary antibodies were used at the following concentrations: α-BiP ([]; rabbit; 1:2,000), α-mNG (ChromoTek; mouse; 1:1,000; 32F6), α-Hb (Abcam; rabbit; 1:2,000; ab92492). Bound primary antibodies were detected by chemiluminescence using horseradish-peroxidase-conjugated secondary antibodies (Agilent; 1:5,000–1:3,000; P039901-2/ P044701-2) and ECL reagent. Co-immunoprecipitation To identify potential PfATG18-interacting proteins, PfATG18 was immunoprecipitated from atg18-mNG:loxP parasites using the mNeonGreen-Trap Magnetic Agarose kit (ChromoTek, ntmak). Briefly, 200-mL cultures at 7% parasitemia and 1% hematocrit were harvested, and parasites were released by treatment with 0.03% saponin. Following centrifugation, parasite pellets were washed with PBS and used as input material for Co-IP. The protocol was performed according to the manufacturer’s instructions with minor modifications. For each Co-IP experiment, pellets were lysed in 800 μL of NP40-based lysis buffer, after which the lysate was diluted 1:2 with dilution buffer and incubated overnight at 4°C with anti-mNG nanobody-coupled magnetic beads under constant agitation. The beads were then washed four times and heated at 95°C for 5 minutes in elution buffer containing 1% SDS and 20 mM TRIS (Carl Roth, 4855.2; pH 7.5). Eluted proteins were collected from the supernatant and stored at -80°C until analysis by liquid chromatography – mass spectrometry, as detailed in the . Use of generative artificial intelligence During the preparation of this manuscript, the authors used ChatGPT (OpenAI), based on the GPT-5.2 large language model (February 2026 version), for language editing. The tool was used solely to refine wording and grammar. All scientific content was developed, reviewed, and approved by the authors, who take full responsibility for the manuscript. Funding Statement This work was funded by the Bundesministerium für Forschung, Technologie und Raumfahrt (BMFTR; Federal Ministry of Research, Technology and Space; grant number 01KI2104, to JMM), providing salary support for YS, MS, and JMM. We also acknowledge financial support from the priority program SPP 2225 “Exit Strategies of Intracellular Pathogens” of the Deutsche Forschungsgemeinschaft (DFG; German Research Foundation; projects 531646514 to JMM and 446605368 to UD) and from the Werner Otto Stiftung (project 3/102, to JMM). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Responsibility for the content of this publication lies solely with the authors. URLs: BMFTR https://www.bmftr.bund.de/DE/Home/home_node.html DFG https://www.dfg.de/ Werner Otto Stiftung https://www.werner-otto-stiftung.de/.