Apicomplexan motility depends on the operation of an endocytic-secretory cycle

Apicomplexan parasites invade host cells in an active process, involving their ability to move by gliding motility and invasion. While the acto-myosin-system of the parasite plays a crucial role in the formation and release of attachment sites during this process, there are still open questions, such as how the force powering motility is generated. In many eukaryotes a secretory-endocytic cycle leads to recycling of receptors (integrins), necessary to form attachment sites, regulation of surface area during motility and generation of retrograde membrane flow. Here we demonstrate that endocytosis operates during gliding motility in Toxoplasma gondii and appears to be crucial for the establishment of retrograde membrane flow, since inhibition of endocytosis blocks retrograde flow and motility. We identified lysophosphatidic acid (LPA) as a potent stimulator of endocytosis and demonstrate that extracellular parasites can efficiently incorporate exogenous material, such as nanogold particles. Furthermore, we show that surface proteins of the parasite are recycled during this process. Interestingly, the endocytic and secretory pathways of the parasite converge, and endocytosed material is subsequently secreted, demonstrating the operation of an endocytic-secretory cycle. Together our data consolidate previous findings and we propose a novel model that reconciles parasite motility with observations in other eukaryotes: the fountain-flow-model for apicomplexan parasite motility.


INTRODUCTION
The intracellular protozoan parasite Toxoplasma gondii infects nearly 2 billion people globally. This apicomplexan can cause severe disease in immunocompromised people and can lead to miscarriage or malformation of the foetus in pregnant women (1). During the acute phase of infection, the tachyzoite rapidly replicates inside the host cell within a specialised compartment, the parasitophorous vacuole, which itself is formed during active invasion (2).
Like all apicomplexans, T. gondii invades host cells in an active process, involving both the parasite's ability to move by gliding motility and invasion factors derived from the unique secretory organelles localised at the parasite's apical pole (micronemes and rhoptries) (3). According to the linear motor model, micronemal transmembrane proteins are secreted at the apical tip of the parasite and act as force transmitters by interacting with both the substrate and the acto-myosin system of the parasite.
However, recent studies suggested a different role of the acto-myosin system during motility, since parasites devoid of F-actin are significantly reduced in overall motility, but still capable of moving at a similar speed as wild-type parasites. This surprising finding has been attributed to a function of the acto-myosin as a surface clutch that allows force transmission by regulating the formation and release of attachment sites, akin to other eukaryotes (4). In good agreement with this hypothesis, mutants for this system cannot efficiently attach to the substrate (5)(6)(7)(8).
During motility, most eukaryotic cells show a capping activity of surface ligands, which is dependent on actin, microtubules, and a secretory-endocytic cycle, leading to the establishment of a retrograde membrane flow (4,9). A recent study on Dictyostelium provided direct evidences for the fluid flow model during cell migration (10). This study demonstrated that, during migration of Dictyostelium, the membrane volume of the cell remains constant due to the occurrence of a secretory-endocytic cycle.
This circulation follows a fountain flow model, where new membrane lipids are delivered to the anterior cell membrane, whereas excess membrane is recycled at the rear. Interestingly, in this study a direct relationship between cell migration and membrane turnover rate was observed, suggesting that the cells establish a fluid drive that contributes to the generation of force required for motility, as suggested previously (4). Importantly, it appears that myosin and actin have only a supporting function in the establishment of the fluid drive, since treatment with actin-or myosin-disrupting drugs, such as latrunculin B or blebbistatin, did not significantly affect membrane movement (10). In good agreement, retrograde membrane flow in apicomplexan parasites is not strictly dependent on parasite actin, as shown for Plasmodium sporozoites and Toxoplasma tachyzoites (7,11).
It is well accepted that gliding motility of apicomplexans depends on the regulated secretion of the apically localised micronemes (12). While this dependency was previously contributed to the secretion of surface ligands, such as MIC2, that are required as force transmitters, it is also possible that polarised secretion is required for the generation of retrograde membrane flow, akin to the fountain flow model (10) and as previously suggested for T. gondii (7,13). However, to date it is not fully understood how apicomplexan parasites ensure a constant cell surface during motility by removing excess membrane deposited on the surface due to microneme secretion.
While the previously described shedding of membrane trails during motility (14) might contribute to a constant membrane content and cell surface, it appears likely that, as suggested by the fountain flow model (10), excess membrane is internalised and recycled during motility. Besides, if shedding was the only mechanism to achieve membrane balance, the parasite would be forced to synthesize an energetically unfeasible quantity of new membrane lipids/proteins, which seems unrealistic for a parasite with limited metabolic abilities.
Here we set out to determine if extracellular parasites are capable of efficiently recycling membrane and taking up exogenous material via endocytosis. To date, uptake of exogenous material has been demonstrated during the intracellular stages of the parasite (15,16). In other eukaryotes, endocytic processes play key roles in membrane dynamics, making it an important participant in cell motility (17,18). Endocytosis can occur via different mechanisms and are roughly defined as CDE (clathrindependent endocytosis) or CIE (clathrin-independent endocytosis) (19). Apicomplexan genomes lack many factors known to be involved in the endocytic system, such as the ESCRT complex, and previous reverse genetic analysis suggested that the remaining factors were repurposed to contribute to the biogenesis and maintenance of unique organelles, such as the Inner Membrane Complex (IMC) or the secretory organelles (20)(21)(22)(23)(24)(25).
Here, we demonstrate the implication of endocytosis in the maintenance of retrograde membrane flow and provide a link between this process and gliding motility, in good agreement with the fountain flow model (4,10). We demonstrate the capacity of extracellular tachyzoites to take up phospholipids, 10 nm nanogold particles, and antibodies directed against parasite surface proteins. Interestingly, endocytic uptake of material appears to occur in a clathrin-independent manner and follows the known secretory pathway of the parasite, with accumulation of material in the rhoptries but also vacuolar-like compartment (VAC, PLV, (26,27)).
Together our data demonstrate the existence of an secretory-endocytic cycle during parasite motility that appears capable of generating the force for motility and therefore fully supports the hypothesis of a fountain-flow model, as suggested for other motile eukaryotic cells (10) .

Fountain flow model and evidence of endocytosis implication in T. gondii motility
The fountain flow model has been recently demonstrated to operate during eukaryotic cell motility, such as in Dictyostelum discoideum (10). This model predicts the establishment of a retrograde membrane flow by localised secretion (at the anterior end of the cell), followed by endocytic recycling to ensure membrane balance. In this respect, apicomplexan parasites are a prime example of highly polarised cells, where the micronemes are secreted at the apical tip during gliding (Fig.1A), which, in analogy to D. discoideum, should result in retrograde membrane flow.
To analyse retrograde membrane flow in T. gondii tachyzoites we previously adapted a translocation assay (28) that follows the transport of beads to the posterior end of the parasite. This 'capping' has been directly implicated in the motility of T. gondii, and, importantly, can occur in the absence of the acto-myosin motor (7). Three phenotypes are observable during this assay: "Unbound" parasites without beads, "Bound" parasites with beads distributed throughout the plasma membrane, and "Capped" parasites that have translocated all the bead to their basal pole (Fig. 1B).
We performed live imaging of the capping process in the presence and absence of different inhibitors or proteins that interfere with the acto-myosin system, secretion, or endocytic processes in other eukaryotes (Fig.1C, Video S1). We incubated parasites with 40 nm latex beads at 4°C to allow binding of the beads to the parasites surface. Upon a temperature shift to 37°C capping occurs rapidly, and beads accumulate at the posterior pole of the parasite within ~25 seconds (Fig. 1C). Importantly, we found that parasite gliding correlates with bead translocation (Video S2), since ~76% of motile parasites showed translocation, while ~24% of motile parasites did not have any beads on the surface, indicating that no initial binding of the beads occurred or that beads were shed after translocation, as seen in movie S3. Importantly, no gliding parasites were identified where beads remained immobile while bound to the parasite plasma membrane (Fig.1B, D, E). Interestingly, parasites can translocate beads without moving, demonstrating that retrograde flow can occur in the absence of gliding, while gliding does not occur in the absence of retrograde flow.
The mechanism underlying parasite membrane balance is unknown and suggested to depend exclusively on membrane shedding and processing of micronemal transmembrane proteins (29). We hypothesised that, akin to other eukaryotes, membrane balance is also ensured by endocytic recycling of excess membrane and proteins (10). To determine if T. gondii retrograde flow could be dependent on a similar mechanism, we tested conditions that inhibit or alter secretion/exocytosis using established inhibitors of endocytosis or parasite strains, such as parasites expressing dominant negative (DN) DrpB (DrpB-DN) (23), where micronemes are absent ( Figure 1F).
Parasites incubated on ice bound beads (92.5±2.6%), but no translocation was observed; a temperature shift to 37°C resulted in translocation in ~27% of parasites. Interestingly, incubation of parasites in the presence of 0.5 µM Cytochalasin D (CD, a drug used to disrupt F-actin) did not result in significant reduction of bead translocation, confirming that retrograde membrane flow can occur in the absence of a functional acto-myosin system, as reported previously (7). In sharp contrast, abrogation of microneme secretion, either by incubation of parasites in endo buffer (2) or induction of the DrpB-DN strain, abolishes bead translocation, demonstrating that retrograde flow depends on polarised secretion as predicted by the fountain flow model ( Fig.1A; (10). To evaluate if capping could depend on endocytosis, we used well established inhibitors of endocytosis, such as Phenylarsine oxide (30) and Trifluoroperazine (31), as well as a DN strain for clathrin heavy chain (CHC-Hub) (21). While the endocytosis inhibitors abrogated capping (2.3±0.33% and 9.8±0.83% capping respectively), expression of dominant negative CHC1 did not result in significant reduction of capping (-Shield: 21.3±3.1%, +Shield: 28.9±2.6% capping). Together, these results suggest that retrograde membrane flow depends on an endocytic mechanism, which might be a form of CIE. In good agreement, to date no clathrin-coated vesicles have been identified at the parasite surface and a previous study did not implicate CHC in endocytosis in T. gondii (21).
Next, we determined the average time required for capping under the same conditions as above ( When motility was analysed using the same conditions, we found a clear correlation between capping and motility. We confirmed previous findings, demonstrating that interference with the acto-myosin system of the parasite results in significantly reduced overall gliding motility (Fig.1H, I). Importantly, parasites still capable of gliding did so at similar speeds as control parasites (Fig. 1J), as described previously (7). In contrast, conditions that resulted in less and/or slower capping resulted in both fewer parasites capable of initiating gliding motility (Fig.1H, I), and parasites moving significantly slower (Fig.1J), when compared to controls.
Together these data suggest a link between parasite motility and capping, and that the parasite is capable of ensuring membrane balance by a secretory-endocytic cycle, as proposed by the fountainflow model (Fig.1A).

Extracellular T. gondii tachyzoites are capable of endocytosis
While it has been recently demonstrated that intracellular parasites are capable of taking up host proteins by an endocytosis-like mechanism (16), to date the presence of endocytic activity of extracellular parasites is still under debate. In order to visualise membrane turnover, we assessed the capacity of T. gondii to take up fluorescent lipids, such as Lysophosphatidyl choline (LPC), Lysophosphatidic acid (LPA), or BODIPY (BP) (Figure 2A). In all cases, we observed efficient uptake of the lipids, characterised by the occurrence of discernible green fluorescent vesicles inside the parasite, that might be associated with the parasite's secretory system (Figure 2A, B). The lipids were taken up at comparable rates (BP:82.9±10%, LPC:66.1±2%, LPA: 67.9±2.3%) and uptake only occurred when parasites were incubated at 37°C, while no uptake was obvious at 4˚C or when dead parasites were incubated with these lipids ( Fig. 2A, B, S1). Together these data demonstrate an active uptake of phospholipids in the majority of extracellular parasites.
Next, we assessed if the observed uptake is due to an endocytosis-like mechanism. In the absence of well-established protein markers for endocytosis, we decided to analyse the uptake of 10 nm nanogold particles (NGP) that are regularly used to analyse endocytosis in other eukaryotes (32).
When wildtype parasites were incubated with NGP it was possible to detect NGP in vesicular structures inside the parasite (Fig.2C, D, S1, S2, Video S4). Interestingly, we found that NGP uptake is significantly stimulated when parasites were incubated with LPA, but not LPC or BODIPY, with up 50% of the parasite population demonstrating NGP uptake in presence of LPA, compared to less than 5% in its absence (Fig.2E, F, S2). NGP uptake was inhibited by incubation at 4°C or when dead parasites were analysed ( Fig.2A), confirming that this represents active uptake rather than passive diffusion of NGP. LPA is a phospholipid that is naturally present in serum (33) and has been reported to be an endocytosis stimulator in mammalian cells (34,35). No toxic effect or alteration in invasion, replication, or parasite morphology was observed in parasites incubated with LPA ( Figure S2B, C, D) or LPC (data not shown). Triggering microneme secretion by addition of 2% ethanol did not lead to any significant increase in NGP uptake (1.1±0.4%, Figure 2E).
We hypothesised that NGP uptake is an endocytic process that can be triggered by LPA. Consequently, we speculated that parasite surface proteins, such as the GPI-anchored surface antigen 1 (SAG1), are likely to be recycled. To test this hypothesis, we incubated parasites in the presence of LPA and α-SAG1. After 30 minutes, parasites were fixed, stripped of unbound α-SAG1, and analysed for uptake of the antibody. We found that treatment of parasites with LPA leads to internalisation of α-SAG1 in ~20% of all parasites (Fig.2G, H), demonstrating the existence of an endocytic recycling pathway.
Importantly, internalised α-SAG1 co-localises with LPA and internalised NGP, demonstrating that α-SAG1, LPA, and NGP utilise the same pathway. However, we can only speculate as to why the observed percentage of parasites recycling α-SAG1 is significantly lower compared to NGP uptake. It is possible that SAG1 is rapidly recycled or processed in the VAC (16).
In contrast to proteinaceous markers such as α-SAG1, the use of a non-reactive particle like NGP therefore allowed us to detect endocytosis more efficiently, and so we chose to continue this study using LPA and NGP to further characterise uptake dynamics and the uptake pathway.

Dynamics of the endocytic cycle.
Next, we analysed uptake of LPA and NGP over time (Fig. 3A, B, S3). Interestingly, intracellular LPA can be detected rapidly (from 17.0±1.6% 1 min after to 67.9±2.3% 30min after addition), while NGP detection lagged behind (0.3±0.3% 1min after to 50.3±2.8 30min after addition, Fig.3A). Though this difference could be explained by a weak signal to noise ratio of single gold beads, requiring accumulation of NGP in endocytic compartments before they can be detected, it is also possible that this lag phase is due to the requirement of LPA to stimulate endocytosis of more bulky material, as described in other eukaryotes (34,35) . Furthermore, the average number of intracellular vesicles increases over time, from ~2 at 5 minutes to ~5 at 30 minutes after addition of LPA (Fig. 3B, S3). If a secretory-endocytic cycle operates within the parasite, it is possible that material that entered via the endocytic route could be recycled and secreted. To test this hypothesis, parasites were pre-treated with LPA and NGP for 30 min, before excess material was washed away, and parasites transferred to new dishes containing minimal media, complete media, or host cells with complete media for 30 min before fixation. We found that NGP and LPA signal changed when parasites were placed into complete media or dishes containing host cells (Fig.3E, F, G). In minimal media, the percentage of positive parasites was similar prior to washing, for both LPA (68.0±3.7% versus 58.9±3.4%) and NGP (42±2.0% versus 33.6±4.3%). In contrast, when the parasites were placed in complete media or in the presence of host cells, a drastic reduction in signal was detected for both LPA (67.0±2.3% vs 32.0±2.4%, Fig. 3E) and NGP (42.9±2.0% vs 8,9±2.8%, Fig 3H). Importantly under these conditions a different distribution of vesicular structures became evident ( Figure 3I, J), where LPA was observed at the apical and basal ends of the parasite, as well as at the parasite surface. LPA was also observed in trails when parasites were incubated with host cells, suggesting secretion of the LPA/NGP-positive vesicles. This is supported by a decreased intensity of LPA signal between attached and invaded parasites (Fig. 3G).
Taken together, these data indicate that a complete endocytic-exocytic cycle is occurring under these conditions.
To investigate the nature of the identified compartments and the origin of uptake in more detail, we performed correlative light and electron microscopy (CLEM). Parasites were incubated with LPA and NGP for 30 minutes before fixation (Fig. 5A). At least two types of vesicles could be identified: vesicles with a similar density as the cytoplasm (Fig. 5A 1 and 2) and other more translucent vesicles (Fig. 5A 3). Next, classical Transmission Electronic Microscopy (TEM) was used (Fig. 5B, C, D, E, G). As observed by IFA and CLEM, NGP were found in different locations within extracellular tachyzoites. NGP accumulated with different densities inside vesicles (Fig.5B, C) and were found in at least 3 types of vesicles: large translucent vesicles (300nm -500nm, Fig. 5C panel 1), medium size dense vesicles (215nm-375nm, Fig.5C2) and small vesicles (80nm-200nm, Fig. 5C panel 3 and 3'). In good agreement with the co-localisation experiments, NGP were frequently observed inside the VAC (Fig. 5D). In some instances, NGP were also detected inside the rhoptry bulb both in TEM and Tomography, confirming the co-localisation with ROP1 observed by IFA, and suggesting that some of the vesicles could fuse or exchange material with the rhoptries (Fig. 5E, F). In contrast, NGP were never seen in dense granules or micronemes. We also identified invaginations at the surface of the parasite that contain NGP, probably representing the point of uptake. These structures are delineated by the plasma membrane, demonstrating that NGP are actively taken up in an endocytic-like process. Importantly, these invaginations are not electron dense and a classical clathrin cage could not be detected, suggesting a clathrin-independent uptake mechanism. (Fig. 5G).
Endocytosis is linked to motility of T. gondii tachyzoites.
As described above (Fig. 1), we were able to observe a link between capping of beads and motility. To determine if a similar correlation between retrograde membrane flow and endocytosis exists, we coincubated parasites with 40nm beads (used in the capping assay) and LPA (Fig. 6A, B, C). The addition of lipids did not significantly increase overall capping activity of T. gondii (Capped -LPA: 30.6±5.8, Capped +LPA 33.0±6.1). Interestingly, we observed that almost all capped parasites (85.6 ± 9.4%) also presented LPA positive vesicles, as illustrated in Figure 6B, while parasites that only bound latex beads to their surface did not show a high percentage of LPA uptake (11,7 ± 8.4%, Fig. 6A, B). This correlation between capping and LPA uptake was also observed using live microscopy (Fig. 6C, Video S5).
To address the importance of endocytosis during gliding motility, parasites were incubated with LPA and NGP and put on coverslips coated with FBS to perform gliding assays. After fixation, parasites were incubated with α-SAG1 antibody to observe trail depositions (Fig. 6D, E, Video S6). We obtained a similar percentage of NGP uptake as in our previous experiments (50.3±6.7%). Strikingly, considering only gliding parasites, i.e. those that deposited trails, NGP uptake markedly increased (82.5±12.5%).
Finally, we used different mutant strains and drugs as above (Fig.1, 6F). Endocytosis inhibitors that impacted both gliding and capping strongly diminished NGP uptake. In addition, phenylarsine did so without impacting LPA uptake (LPA: 67.3±6.2%, NGP: 6.5±2.4%), whereas TFDC inhibited both LPA and NGP uptake (LPA: 6.8±7.4%, NGP: 0.6±0.6%). We also tested the DN CHC strain, which does not significantly affect either gliding or capping. We observed a slightly reduced uptake of NGP (Ctrl: 48.5±7.1% vs DN:24.6±1.7), whereas no reduction in LPA uptake was observed. This suggests that CHC is not critical for endocytosis and that the observed reduction of NGP uptake might be due to downstream effects caused by blocking post-Golgi transport through CHC-DN expression (21).
Taken together, these results demonstrate a clear link between retrograde membrane flow, endocytosis, and parasite gliding motility, and therefore fully support the fountain flow model.

A link between gliding motility, retrograde membrane flow, secretion, and endocytosis
Apicomplexan gliding motility has been considered a "unique" form of cell motility that depends on special secretory organelles found in apicomplexans, which secrete transmembrane proteins to link the acto-myosin system anchored in the IMC of the parasite to the substrate during motility. The unusual features of parasite actin, combined with the absence of direct homologues of many classical actin binding partners, led to the hypothesis that apicomplexan motility is distinct from other classical eukaryotic mechanisms. Here we argue that this form of motility is similar to amoeboid motility and that many of the identified crucial elements for gliding motility play analogous roles compared to factors found in other eukaryotes. A good example are micronemal proteins, which act similar to integrins and often contain integrin-like extracellular domains, such as MIC2 (36). Micronemal proteins are required for attachment to the surface and have been recently shown to interact with Factin via the glideosome-associated connector (GAC) (37), which appears to fulfil an analogous role to talin, which connects integrins to F-actin in motile eukaryotic cells (38). Interestingly, and very similar to findings in apicomplexan parasites, recent results in motile eukaryotic cells have led to a reassessment of the traditional motility models. Of note, analysis of cell migration in a 3Denvironment, which is arguably more physiological than a 2D-environment, has led to a rich body of literature describing new motility mechanisms such as blebbing, osmotic engines, fountain-flow, membrane stretch, etc. (10,(39)(40)(41)(42)(43). Although lengthy description of these various models is outside the scope of this discussion, it is important to note that motile cells can move by employing more than one mechanism depending on the environmental conditions encountered.
In the case of apicomplexan parasites, the current model describes a simple linear motor that is required for all motility processes of the parasite, including host cell invasion (29). However, the Importantly, retrograde membrane flow is implicated in many different motility modes of eukaryotic cells and recent evidence demonstrates important roles for membrane trafficking in the regulation of cell migration in a variety of contexts. For example, one critical role of an endocytic-secretory cycle is the internalisation and recycling of adhesion receptors, such as integrins or syndecans (50). Another critical role is the maintenance of a constant cell surface (51,52). Indeed, recent studies suggested the operation of a so-called fountain-flow model that predicts that surface area regulation and retrograde membrane flow is caused by a secretory-endocytic cycle (10). Here we demonstrated that the situation in the apicomplexan parasite is very similar and fully supports the fountain-flow model proposed previously for other eukaryotic cells (53). Previous studies implicated the generation of a retrograde membrane flow in parasite motility (7,11,54). Interestingly, membrane flow itself is relatively resistant to actin-modulating drugs, whereas force generation depends on a functional actomyosin system. Here, we extended our analysis of retrograde membrane flow using a bead translocation assay (7) and demonstrate a strong link between the occurrence of bead translocation, parasite motility, and the operation of an endocytic-secretory cycle. We verified that interference with the acto-myosin system of the parasite has little effect on bead translocation, while blocking apical microneme secretion or endocytosis abrogates bead translocation and parasite motility (Fig. 1F, G, H,

Extracellular parasites can take up exogenous material and recycle surface proteins
Uptake of host cell material from intracellular parasites has recently been demonstrated and it appears that this material is taken up by an endocytic route that merges with the secretory pathway of the parasite (15). This endocytic process appears to occur rapidly, with endocytosed proteins eventually reaching the vacuolar-like compartment (26,27) of the parasite, in which they are digested (16). However, it was previously unclear whether extracellular parasites are capable of similar uptake, especially since a number of trafficking factors normally involved in endocytosis have been shown to be required for the transport of material to unique secretory organelles, leading to the hypothesis that endocytic factors have been repurposed during apicomplexan evolution (55). Here we demonstrate that extracellular parasites, like intracellular parasites, are capable of endocytosis, and that the parasite appears to recycle surface proteins, such as SAG1. Interestingly, we observe that lipid dyes are efficiently taken up, but only LPA appears to stimulate endocytosis of larger material, such as 10nm gold beads. It is thus likely that LPA triggers endocytosis via an as yet unknown signalling cascade. LPA has been demonstrated in numerous systems to activate endocytic uptake (34,35), and can be converted to phosphatidic acid (PA) (56), which can act as a second messenger. For example it can promote both CME and CIE (57,58), and can also lead to the recruitment and activation of proteins directly involved in the trafficking machinery (59). Interestingly, phosphatidic acid (PA) has been previously implicated in regulating microneme secretion and invasion in apicomplexan parasites (60), making it attractive to speculate that PA is a central mediator for the secretory-endocytic cycle.
Stimulation of parasites with LPA leads to efficient uptake of NGP, which co-localise with LPA-positive vesicular structures. Importantly, we found that these structures co-localise with markers of the endomembrane system of the parasite, such as the ER, Golgi, VAC, and rhoptries, with a clear accumulation in the VAC, suggesting that the same system is employed as for the uptake of host cell material during intracellular stages (16). Ultrastructural analysis confirms these findings and suggests that the material is taken up at surface invaginations that appear to be distinct from the micropore of the parasite (61). To date, our attempts to obtain mechanistic insight regarding the endocytic process have been unsuccessful. While established endocytosis inhibitors show the expected effects, i.e. marked decrease in uptake of material, analysis of dominant negative mutants for clathrin or dynamins had no significant effect on endocytosis. Furthermore, no clathrin-coated pits, a hallmark of CME, could be detected in our ultrastructural analysis. While this might suggest that the analysed factors are not involved in endocytosis, it is also possible that the conditional mutants used do not have the right kinetics for downregulation of the respective genes, since this might already cause parasite death within the host cell, before extracellular endocytosis can be analysed. Therefore, faster conditional regulation systems should be used in future studies to re-analyse these factors, such as the auxin-inducible degron system (62).
We demonstrate here that T. gondii is capable of performing endocytosis in the extracellular stage by presenting a method to visualise internalization and recycling of both surface proteins and exogenous material (SAG1 and LPA/NGP, respectively). It is interesting that NGP were found in the rhoptries, but not micronemes, since components of both organelles are delivered by a common secretory system (63). One possibility is that NGP accumulate in the rhoptries, since these are not secreted in extracellular parasites, while micronemes can be secreted (for example during motility). Indeed, we demonstrate here that the NGP are secreted upon incubation of parasites in complete media or on host cells, providing evidence for the existence of an endocytic-recycling pathway that operates in stimulated parasites.

The Fountain-Flow-Model for apicomplexan parasites
Based on these data, and previous data that suggested an important role of the acto-myosin-system as a clutch for the transmission of force (7), we suggest that the generation of retrograde membrane flow is critical for gliding motility. This flow can be generated by the acto-myosin-system of the parasite as well as regulated secretion and endocytosis, as suggested for other eukaryotes (53). According to this model, apical microneme secretion needs to be balanced by endocytic recycling, resulting in a lipid drive or fountain flow (10). In good agreement with this model (Figure 7), we demonstrate that gliding parasites generate retrograde membrane flow while simultaneously internalizing NGP; furthermore, inhibitors of uptake also block retrograde membrane flow and gliding.
In contrast, interference with the acto-myosin system does not cause a block in retrograde membrane flow or uptake, resulting in parasites that glide less, which can be attributed to less surface attachment, as previously shown (7,8)

Outlook and Summary
While this study reconciles many previously conflicting data, it does not directly address the mechanisms involved in endocytic uptake, as discussed above. It will now be important to identify the crucial trafficking factors directly involved in endocytosis. It is likely that the parasite evolved specific structures for endocytic uptake, as suggested by the presence of the invagination seen in TEM analysis.
A recent genome wide screen (64) demonstrated the essentiality of hundreds of hypothetical genes.
Many of them might well be involved in an essential uptake pathway, required for gliding motility and host cell invasion. Therefore, with the establishment of a reliable uptake assay, it is now possible to phenotypically screen for essential genes involved in this process, which will not only result in a fundamental understanding of this process, but also in the identification of novel intervention strategies.

Materials and Methods:
Cloning DNA constructs: All primers used in this study are listed in Table 2 and were synthesised from Eurofins (UK).

T. gondii transfection and selection:
To generate stable parasite lines, 1x10 7 freshly lysed RH ∆hxgprt or RH-DiCre ∆ku80 parasites were transfected with 20 µg of DNA by AMAXA electroporation. Drug selection was carried out with either mycophenolic acid and xanthine as described in (65), or with bleomycin.
The resultant transfectants were selected for clonal lines expressing VPS35-HA or VPS53-HA in presence of 25 μg/ml mycophenolic acid and 40 μg/ml xanthine and subsequently cloned by limiting dilution. Specific integration was confirmed by analytical PCR on genomic DNA using primers upstream the homology region inserted in the LIC vector, and a reverse primer binding the LIC HA region ( Table   2). Inducing conditional knockdown lines and protein expression: dd-DrpBDN , dd-CHCDN,, dd-DrpBDN,, dd-Rab18-myc, dd-Rab4-myc and dd-RAB2-myc parasites (Table 1) were grown until the vacuoles were ready to lyse. Shield was added 6 hours prior to the parasites being used for experiments. act1 cKO, mlc1 cKO were induced as previously described (7).

Phenotypic characterisations
Trail deposition assay: Gliding assays were performed as described before (7). Briefly, freshly lysed parasites were allowed to glide on FBS-coated glass slides for 30 min before they were fixed with 4 % PFA and stained with α-SAG1 under non-permeabilising conditions. Mean values of three independent experiments +/-SEM were determined. Where drugs were used, parasites were preincubated for 10 minutes in the respective concentration before the start of the assay: 0.5 μM Cytochalasin-D (CD) (Sigma), 10 µM phenyl arsine oxide (Sigma) or 50 µM Trifluoperazine dihydrochloride (TFDC) (Sigma). The same concentrations were used in the different assays.
2D motility assay: Time-lapse video microscopy was used to analyse the kinetics over a 2D surface similar as previously described (48). Briefly Ibidi μ-dish 35mm-high were coated in 100 % FBS for 2 hours at room temperature. Freshly egressed parasites were added to the dish. Time-lapse videos were taken with a 40X objective at 1 frame per second using a DeltaVision ® Core microscope. Analysis was made using ImageJ wrMTrckr tracking plugin. For analysis, 20 parasites were tracked during both helical and circular trails with the corresponding distance travelled, average and maximum speeds determined. Mean values of three independent experiments +/-SD were determined.
Invasion assay: For the assay, 5x10 4 freshly lysed parasites were allowed to invade a confluent layer of HFFs for 1 hour after 30 minutes treatment with or without LPA. Subsequently, five washing steps were performed for removal of extracellular parasites. Cells were then incubated for a further 24 hours before fixation with 4% PFA. Afterwards, parasites were stained with the α-IMC1 antibody (8).
The number of vacuoles in 15 fields of view was counted. Mean values of three independent experiments +/-SEM were determined.
Capping assays: Capping assays were performed as previously described (7 Live capping assays: Capping assay were adapted for live microscopy. Parasites were prepared as described above. After the addition of the diluted beads (5 µl of beads in 250 µl of H-H buffer) to the parasites, the dish was incubated for 10 minutes on ice. After incubation the media was exchanged for 500 µl of ice cold H-H buffer without beads. The dish was then directly transfer to the microscope.
Time-lapse videos were taken with a 60X objective at 1 frame per second using a DeltaVision ® Core microscope. Analysis was made using ImageJ. Immunofluorescence analysis: Carried out as previously described (8). Briefly, parasites were fixed in 4% paraformaldehyde for 10 min at 4˚C. Afterwards, coverslips were blocked and permeabilised in 2% Samples were processed for routine electron microscopy as described previously (49) and examined in a JEOL 1200EX electron microscope.
Correlative light-electron microscopy (CLEM): Uptake assays were carried out in gridded glass bottom petri dishes (MatTek). Parasites presenting clear LPA and NGP uptake were imaged with SR-SIM in an ELYRA PS.1 microscope (Carl Zeiss, Germany), and the material was fixed in 2.5% glutaraldehyde and 4% paraformaldehyde in 0.1 M phosphate buffer; and processed for transmission electron microscopy as described previously (49). Thin sections of the same areas imaged in 3D-SIM were imaged in a Tecnai T20 transmission electron microscope (FEI, Netherlands).