Characterisation of TbSmee1 suggests endocytosis allows surface-bound cargo to enter the trypanosome flagellar pocket

The flagellated parasitic protist Trypanosoma brucei lives in the bloodstream of its mammalian hosts, in continuous exposure to the immune system. Endocytosis and exocytosis are used to remove bound antibodies from the cell surface and scavenge macromolecular nutrients from the surroundings (Borst and Fairlamb, 1998; Engstler et al., 2007; Webster et al., 1990). Remarkably, all endocytosis and exocytosis in T. brucei occur at just a single subdomain of the plasma membrane (Engstler et al., 2004; Grunfelder et al., 2003). This subdomain is a small vase-shaped invagination, called the flagellar pocket, that houses the root of the single flagellum of the cell (Halliday et al., 2021; Lacomble et al., 2009).

Molecules that enter the flagellar pocket, either in the fluid phase or when bound to the surface of the parasite (hereafter ‘surface-bound’), are rapidly internalised by clathrin-mediated endocytosis and then routed to the endosomal/lysosomal system (Link et al., 2021; Overath and Engstler, 2004). Depletion of clathrin results in a gross enlargement of the flagellar pocket due to an imbalance between endocytosis and exocytosis, and causes cargo accumulation inside the flagellar pocket (Allen et al., 2003). Although endocytosis of fluid phase and surface-bound cargo is relatively well characterised in T. brucei, the exact mechanisms by which cargo initially enters the flagellar pocket are less well-understood.

Coiled around the neck of the flagellar pocket on its cytoplasmic face are several poorly characterised cytoskeleton-associated complexes, which might contribute to these processes (Esson et al., 2012; Halliday et al., 2019) (Fig. 1A). The flagellar pocket collar is a multiprotein complex shaped like a cuff bracelet and demarcates the boundary between the flagellar pocket and the flagellar pocket neck. It contains the protein T. brucei (Tb)BILBO1, and a number of its other components such as FPC4 and BILBO2 have recently been characterised (Albisetti et al., 2017; Bonhivers et al., 2008; Florimond et al., 2015; Isch et al., 2021). The centrin arm is a centrin-containing kinked rod which appears to be involved in the biogenesis of various cytoskeleton-associated structures (Selvapandiyan et al., 2007; Shi et al., 2008). The hook complex is a hook-shaped multiprotein complex that sits atop the flagellar pocket collar and alongside the centrin arm. Characterised components of the hook complex include the proteins TbMORN1, BOH1, BOH2 and Bhalin (Broster Reix et al., 2021; Morriswood et al., 2009; Pham et al., 2020; Pham et al., 2019).

Previous characterisation of TbMORN1 revealed that it is essential for the viability of bloodstream form T. brucei cells (Morriswood and Schmidt, 2015). Cells depleted of TbMORN1 have enlarged flagellar pockets, indicative of an endocytosis defect. In addition, although small endocytic reporters such as 10 kDa dextran (hydrodynamic radius ∼1.86 nm) were still capable of entering the enlarged flagellar pocket, the access of larger macromolecules, such as the lectin concanavalin A (ConA, hydrodynamic radius ∼4.2 nm) or bovine serum albumin (BSA, hydrodynamic radius ∼3.51 nm) conjugated to 5 nm gold particles, was blocked (Ahmad et al., 2007; Armstrong et al., 2004; Morriswood and Schmidt, 2015). This phenotype was also observed upon depletion of the hook complex protein Bhalin (Broster Reix et al., 2021). On this basis, it was proposed that the hook complex might regulate the passage of large macromolecules through the flagellar pocket neck, and thereby mediate cargo entry into the flagellar pocket (Morriswood and Schmidt, 2015).

In this study, the hook complex protein TbSmee1 was characterised in bloodstream form T. brucei. TbSmee1 (Tb927.10.8820) was initially identified in a proximity-labelling screen using TbMORN1 as bait and was shown to localise to the shank part of the hook complex (Morriswood et al., 2013). It was subsequently characterised in procyclic form T. brucei, the life cycle stage found within the tsetse fly vector (Perry et al., 2018). In bloodstream form T. brucei, depletion of TbSmee1 phenocopied depletion of TbMORN1 and Bhalin, with ConA, BSA and surface-bound antibodies unable to access the enlarged flagellar pocket. Surprisingly, clathrin-depleted cells also exhibited this phenotype of impaired flagellar pocket access, suggesting that the observed effects following depletion of the hook complex proteins are due to impaired endocytosis and, therefore, that endocytosis is required for the entry of surface-bound cargo into the flagellar pocket.

Bioinformatic analysis predicted that TbSmee1 is composed of three structured domains separated by linker regions (Fig. 1B). A multiple sequence alignment of the TbSmee1 protein with orthologues from other trypanosome species indicated the presence of three blocks of highly conserved sequence, corresponding to regions of predicted secondary structure (Fig. S1). Three-dimensional structural prediction by AlphaFold suggested a close association between the first two folded domains (https://alphafold.ebi.ac.uk/entry/Q38A92) (Jumper et al., 2021; Wheeler, 2021).

Two polyclonal antibodies (labelled 303 and 304) were raised against a recombinant TbSmee1(1–400) truncation, and a third (labelled 508) was raised against two peptides (Fig. 1B, green lines). The specificity of the antibodies was confirmed using immunoblots of whole-cell lysates obtained from TbSmee1 RNAi cells and 3×Ty1–TbSmee1 endogenous replacement cells. In situ tagging of the SMEE1 gene in the 3×Ty1–TbSmee1 cells was confirmed by PCR analysis of genomic DNA (Fig. S2A,B). All three antibodies detected a protein of ∼85 kDa in the immunoblots (Fig. 1C, white arrows). This protein was strongly depleted in the tetracycline-induced RNAi (+Tet) samples (Fig. 1C). An extra band corresponding to 3×Ty1–TbSmee1 was seen in lysates from the endogenous replacement (‘End. repl.’) cells (Fig. 1C). This extra band was also detected using anti-Ty1 (BB2) antibody, confirming its identity as Ty1–TbSmee1 (Fig. 1C, lower panels).

To investigate TbSmee1 association with the cytoskeleton, cells were fractionated using non-ionic detergent (Fig. 1D). The fractions were analysed by immunoblotting, using an antibody recognising PFR1 and PFR2 (hereafter PFR1/2) as markers for the cytoskeleton, and the endoplasmic reticulum chaperone protein BiP as a marker for the cytoplasm. TbSmee1 co-fractionated with PFR1/2, confirming that it was associated with the cytoskeleton (Fig. 1E). Interestingly, the fuzzy appearance of the TbSmee1 band observed in immunoblots of whole-cell lysates was reduced or absent in the fractionation blots (compare Fig. 1C and Fig. 1E). TbSmee1 is heavily phosphorylated in vivo, and is a substrate and potential binding partner of the mitotic kinase TbPLK (Benz and Urbaniak, 2019; McAllaster et al., 2015; Nett et al., 2009; Urbaniak et al., 2013). The existence of different phosphoforms of a protein is known to cause a fuzzy appearance of bands in gels, so it was possible that exposure of TbSmee1 to endogenous phosphatases during the extraction step caused dephosphorylation.

To investigate whether the band collapse could be attributed to dephosphorylation of TbSmee1, extracted cytoskeletons were incubated with exogenous phosphatase at various timepoints prior to immunoblotting. Incubation with exogenous phosphatase resulted in a progressive collapse of the TbSmee1 band over a 20 min period (Fig. 1F). This band collapse was not seen when the samples were in the presence of phosphatase inhibitors either at room temperature (RT) or on ice, or kept on ice (Fig. 1F, last three lanes). The fuzzy appearance of TbSmee1 in immunoblots could therefore be attributed exclusively to phosphorylation.

The localisation of endogenous and tagged TbSmee1 protein was analysed in bloodstream form T. brucei cells using immunofluorescence microscopy (Fig. 1G). All three main stages of the cell division cycle (1K1N, 2K1N and 2K2N, where ‘K’ indicates the number of kinetoplasts and ‘N’ the number of nuclei) were analysed. In 1K1N cells (i.e. those with a single kinetoplast and nucleus), TbSmee1 localised to a single structure near the flagellum base, consistent with the position of the hook complex (Fig. 1G, arrow). In 2K1N and 2K2N cells, there was a third subpopulation of TbSmee1 at varying distances along the cell body in addition to the replicated hook complex (Fig. 1G, arrowheads). The same TbSmee1 distributions were seen using all three anti-TbSmee1 antibodies; anti-Ty1 antibody labelling of 3×Ty1–TbSmee1 cells also produced the same labelling pattern (Fig. S2C,D).

In colabelling experiments, TbSmee1 strongly overlapped with the shank part of the hook complex as labelled by TbMORN1 and TbLRRP1 (Fig. S3A,B). TbSmee1 also strongly overlapped with the hook complex protein Tb927.10.3010. This 133 kDa protein was one of the top hits in the BioID screen using TbMORN1 (Morriswood et al., 2013). Consistent with the convention started with TbSmee1, Tb927.10.3010 was named TbStarkey1, after another of Captain Hook's pirates. Two anti-peptide antibodies were generated against TbStarkey1, and their specificity was confirmed by immunoblotting and immunofluorescence imaging (Fig. S4A,B). Strong overlap was seen between TbSmee1 and TbStarkey1 on the shank part of the hook complex (Fig. S4C). As expected, TbSmee1 showed only a partial overlap with TbCentrin4, a marker of the centrin arm, and little overlap with TbBILBO1, a marker of the flagellar pocket collar (Fig. S3C,D).

To quantify the degree of colocalisation between TbSmee1 and the various hook complex and flagellar pocket collar marker proteins, correlation coefficients from the pairwise labelling experiments were calculated. Moderate correlation was seen among TbSmee1, TbMORN1 and TbLRRP1, with lower values being measured for TbCentrin4 and TbBILBO1 (Fig. S3E).

The flagellum attachment zone (FAZ) is a cytoskeleton-associated apparatus that adheres the flagellum of T. brucei to the cell body (Sunter and Gull, 2016). During replication, a new FAZ is assembled and grows from its initiation point very close to the hook complex towards the anterior end of the cell (Sunter et al., 2015b; Zhou et al., 2015). It was previously shown that a tagged version of TbSmee1 is transiently localised to the tip of the new FAZ in replicating insect-stage T. brucei cells, in addition to its hook complex localisation (Perry et al., 2018).

Consistent with this, endogenous TbSmee1 showed a partial overlap with the posterior end of the FAZ (Fig. 2A, arrow). The extra TbSmee1 subpopulation in 2K1N and 2K2N cells overlapped with the tip of the new FAZ (Fig. 2A, arrowheads).

Using structured illumination microscopy (SIM), TbSmee1 was observed to be slightly in front of the tip of the new FAZ filament (Fig. 2B). This suggested that TbSmee1 might be present at the ‘groove’. The groove, which can be detected using the monoclonal antibody DOT1, is a structure involved in remodelling of the microtubule cytoskeleton during cell replication in bloodstream form T. brucei (Hughes et al., 2013; Smithson et al., 2022). TbSmee1 at the FAZ tip appeared to be enveloped by the DOT1 labelling (Fig. 2C). In summary, in replicating cells, TbSmee1 is present at the groove in addition to the hook complex, and travels in front of the newly assembling FAZ filament.

To determine what parts of the primary structure of TbSmee1 are responsible for targeting to the hook complex, a series of truncation constructs based on the predicted domain architecture of TbSmee1 were designed (Fig. 3A). Cell lines were generated that could inducibly express each of these truncations with an N-terminal Ty1 tag. The presence of the ectopic transgenes in the genomes of these cells was confirmed by PCR analysis of genomic DNA (Fig. S5A). To determine the localisation of each construct and whether it could associate with the cytoskeleton, detergent-extracted cells were analysed using immunofluorescence microscopy. It should be noted that both endogenous SMEE1 alleles were still present, meaning that targeting of the truncation constructs was being assayed in the presence of the endogenous Smee1 protein.

TbSmee1[amino acids (aa) 161–766] correctly localised to the hook complex, indicating that domain 1 is not necessary for targeting (Fig. 3B). Of note, no dominant-negative effects on cell growth were observed upon overexpression of the TbSmee1(161–766) construct. TbSmee1(265–766) also localised correctly, indicating that the predicted linker region between domains 1 and 2 is not necessary for localisation (Fig. 3B). No other truncations were observed to localise to the hook complex. Expression of all TbSmee1 truncations was confirmed by immunoblotting with anti-TbSmee1 and anti-Ty1 antibodies (Fig. 3C; Fig. S5A). Quantification of the immunoblots indicated that most were expressed at around the level of the endogenous protein (Fig. 3D).

Interestingly, TbSmee1(161–766) also localised to the FAZ tip in replicating cells (Fig. S5B). TbSmee1(265–766) did not localise to the FAZ tip, suggesting that the linker region between domains 1 and 2 (aa 161–264) might be required for targeting to this structure. In support of this hypothesis, TbSmee1(2–400) was found to localise to the FAZ tip despite it not localising to the hook complex (Fig. S5C).

The effects of TbSmee1 depletion were analysed using tetracycline-inducible RNAi. Induction of RNAi resulted in a rapid cessation of population growth after around 24 h (Fig. 4A). Visual inspection of the stalled populations at 48 and 72 h post induction showed widespread lysis, confirming that TbSmee1 is essential for the viability of bloodstream form cells in vitro. A shorter time course with higher sampling frequency indicated that the growth defect began after around 20 h of RNAi, and was already clear at 24 h (Fig. 4A, inset).

TbSmee1 depletion was confirmed at the single-cell level using immunofluorescence microscopy (Fig. 4B). TbSmee1 signal was lost from both the hook complex and FAZ tip. The kinetics of TbSmee1 protein depletion on either side of the onset of the growth defect were assessed by immunoblotting. TbSmee1 protein levels were reduced to ∼20–25% at 15 and 18 h post induction, with a further reduction to around 10–15% at 21 h and onwards (Fig. 4C).

The effect of TbSmee1 knockdown on cell cycle progression was assessed by quantifying the numbers of 1K1N, 2K1N, 2K2N, and abnormal cell types in the same time window (Fig. S6A). An increase in 2K1N and a decrease in 1K1N cells was visible from 21 h post induction, followed by an increase in 2K2N cells at the 24 h timepoint. This indicated that cell cycle progression was inhibited from around 21 h after induction of RNAi, correlating with the onset of the growth defect. To summarise, TbSmee1 protein was already significantly depleted at 15 h, with subsequent effects on population growth and cell cycle progression visible from around 21 h.

The effects of TbSmee1 depletion on a panel of hook complex proteins and components of the centrin arm and flagellar pocket collar were systematically evaluated (Fig. S7A–E). TbSmee1 depletion for 24 h did not result in observable effects on the expression levels or localisation of any of the candidates or on flagellum attachment. Conversely, depletion of TbMORN1 for just 16 h resulted in reductions in the levels of TbSmee1 and TbStarkey1 (Fig. S7B,F–H).

To determine the ultrastructural changes caused by TbSmee1 depletion, the cells were imaged by electron microscopy after high-pressure freezing, which gives better morphological preservation than chemically fixing the cells in the growth medium. The most obvious morphological effect of TbSmee1 depletion was the gradual accumulation of cells with enlarged flagellar pockets (Fig. 4D). This presumably was the cause of the lethality phenotype, as depletion of TbMORN1 was also shown to result in progressive enlargement of the flagellar pocket, until cells rounded up and lysed (Morriswood and Schmidt, 2015). Depletion of TbStarkey1 also frequently resulted in the generation of cells with enlarged flagellar pockets, despite causing no growth defect (Fig. S6B–D). These findings, together with the recently published characterisation of Bhalin (Broster Reix et al., 2021), show that depletion of four separate hook complex proteins – TbMORN1, TbSmee1, TbStarkey1 and Bhalin – results in flagellar pocket enlargement, though with varying magnitudes of effect.

Although flagellar pocket enlargement appears to be a consistent phenotype resulting from hook complex protein depletion, it is not an uncommon phenotype in bloodstream form RNAi cells. Importantly, it can be either a direct or an indirect consequence of protein depletion (Ali et al., 2014; Allen et al., 2003; Hall et al., 2004; Hall et al., 2005; Price et al., 2007).

To determine whether flagellar pocket enlargement was an early, and therefore probably direct, consequence of TbSmee1 depletion (i.e. occurring very soon after or even before the onset of the growth defect), flagellar pocket size was assayed using a fluorescent, fixable 10 kDa dextran reporter. Dextran is a polysaccharide that traffics in the fluid phase and is well established as an endocytic marker. Control (−Tet) and TbSmee1-depleted (+Tet) cells were incubated on ice in order to block endocytosis (Brickman et al., 1995). The cells were then incubated with the labelled dextran for 15 min to allow it to enter the flagellar pocket, and fixed afterwards (Fig. 5A). The magnitude of the dextran signal should therefore be proportional to flagellar pocket volume. Visual analysis of the cells confirmed that dextran visibly labelled the flagellar pocket, with a much greater signal seen for cells with an enlarged flagellar pocket (Fig. 5B).

Flow cytometry was then used for high-throughput, unbiased and quantitative analysis of the cells (Fig. 5C). At 15 h post induction, no difference between TbSmee1-depleted cells and controls was observed. At 18 and 21 h post induction, a slight ‘shoulder’ on the +Tet traces became visible, indicating the emergence of cells with higher fluorescence values than in controls. By 24 h post induction, there was a clear hump visible in the traces, indicating a subpopulation with fluorescence values that were sometimes two orders of magnitude greater than those in control cells (Fig. 5C, arrow). Quantification of the flow cytometry data from multiple experiments showed that average fluorescence intensity of the whole +Tet population was noticeably higher than controls at 21 and 24 h post induction (Fig. 5D). Even at 18 h post-induction, i.e. before the onset of the growth defect, there was already a clear increase in the average fluorescence intensity in the +Tet population (Fig. 5D, blue arrow). This strongly suggests that flagellar pocket enlargement is an early and probably direct consequence of TbSmee1 depletion.

It was previously shown that the ability of large cargo to enter the flagellar pocket is affected after knockdown of the hook complex protein TbMORN1 (Morriswood and Schmidt, 2015). Specifically, the fluid-phase marker 10 kDa dextran accumulates in the enlarged flagellar pocket of TbMORN1-depleted cells, whereas larger fluid-phase cargo such as BSA-5 nm gold and large surface-bound cargo such as ConA (which binds to surface glycoproteins) do not access the flagellar pocket lumen.

To test whether TbSmee1 knockdown also impairs flagellar pocket access, TbSmee1 RNAi cells were incubated with both dextran and ConA to simultaneously monitor the uptake of fluid-phase and surface-bound cargo. The cells were first incubated on ice to block endocytosis, and then with the reporters (also on ice) to allow ConA to bind and dextran to enter the flagellar pocket. The cells were then shifted to 37°C to reactivate endocytosis, and subsequently fixed and imaged (Fig. 6A).

In control (−Tet) cells, both dextran and ConA strongly overlapped in the part of the cell corresponding to the endosomal/lysosomal system (Fig. 6B, −Tet, arrow). To confirm that the internalised material was being trafficked to the lysosome, the cells were labelled with antibodies specific for the lysosomal enzyme p67 (Kelley et al., 1999). The dextran reporter was not compatible with immunolabelling, but the ConA signal clearly overlapped with the lysosome marker p67 (Fig. S8, −Tet cells). In TbSmee1-depleted (+Tet) cells, dextran filled the enlarged flagellar pocket, whereas ConA was restricted to one or two small foci that appeared to be on the cell surface (Fig. 6B, +Tet, arrow). No overlap was seen between the dextran and ConA labels, suggesting that ConA was not able to enter the flagellar pocket. In addition, no overlap was observed between ConA and the lysosome marker p67 (Fig. S8, +Tet cells).

To quantify these observations, the cells were grouped into four categories: (1) complete overlap between the two labels; (2) partial overlap between the two labels; (3) no overlap between the two labels; and (4) whole-cell labelling. Whole-cell labelling occurs when the cell has lost integrity, and labelling is found throughout the cytoplasm. Control (−Tet) cells all showed complete overlap between the dextran and ConA, whereas cells depleted of TbSmee1 for 24 h showed >25% of cells with partial or no overlap between the cargoes (Fig. 6C).

The experiments were repeated using fluorescently labelled BSA (Fig. 6D). BSA traffics in the fluid phase and is a physiological cargo, unlike ConA (Coppens et al., 1987). As expected, in control cells, there was strong overlap between dextran and BSA from the endosomal/lysosomal system (Fig. 6E, −Tet, arrow). In TbSmee1-depleted cells, there was once again no overlap between the reporters, indicating that BSA was unable to access the enlarged flagellar pocket. Surprisingly, despite BSA being fluid-phase cargo, there was a punctate signal analogous to that seen with ConA, suggesting that it could at least partly bind to the cell surface (Fig. 6E, +Tet, arrow). Quantification of the data showed that the effect on BSA was more pronounced than that seen for ConA, with >55% of TbSmee1-depleted cells showing no or only partial overlap between the two reporters after 24 h of RNAi (Fig. 6F).

The inability of surface-bound cargo to enter the flagellar pocket of TbMORN1- and TbSmee1-depleted cells could either be due to a defect in endocytosis (indicated by the enlargement of the flagellar pocket), or a secondary effect. To distinguish between these possibilities, the assays were repeated following clathrin depletion (Fig. 7A). Clathrin is an essential endocytic coat protein and its loss results in a block in endocytosis. This results in the expected enlargement of the flagellar pocket and ultimately cell lysis, and, in fact, this phenotype was first described as a result of clathrin depletion (Allen et al., 2003). In control cells, as expected, there was strong overlap between the dextran and ConA cargoes in the region of the endolysosomal system (Fig. 7B, −Tet, arrow). In clathrin-depleted cells, however, there was again no overlap observed between the two reporters, and ConA did not appear able to access the enlarged flagellar pocket (Fig. 7B, +Tet, arrow). Quantification of the data showed that >50% of clathrin-depleted cells showed either partial or no overlap between the two reporters after 19 h of RNAi (Fig. 7C). The same effect was seen in assays using dextran and BSA, with little to no overlap between the reporters in clathrin-depleted cells (Fig. 7D–F). These results indicated that inhibition of endocytosis by itself resulted in the failure of either ConA or BSA to enter the enlarged flagellar pocket.

One of the main functions of endocytosis in trypanosomes is to remove any antibodies bound to the surface glycoprotein coat. To determine whether the effects observed above using ConA and BSA reporters also apply to antibodies, antibody uptake assays were carried out. As the uptake of surface-bound antibodies occurs in a matter of seconds, a slightly modified assay protocol was used (Fig. 8A). The cells were first incubated on ice to block endocytosis, and then incubated with antiserum specific for the variant surface glycoprotein (VSG), which accounts for the overwhelming majority of the surface glycoprotein coat. A sample of cells were fixed at this timepoint (t=0) to confirm surface binding of the antibodies. The remaining cells were washed (at low temperature), after which a second sample was fixed (t=1). The remaining cells were shifted to 37°C to reactivate endocytosis and incubated for 2 min to allow the antibodies to enter the flagellar pocket and then be internalised by endocytosis.

The distribution of anti-VSG antibodies in control and TbSmee1-depleted cells was analysed using microscopy (Fig. 8B). In control (−Tet) cells, anti-VSG signals were observed at the cell surface (t=0), then at the location of the flagellar pocket (t=1) and finally at the endosomal/lysosomal system after allowing endocytosis (t=2) (arrows, upper panels). In TbSmee1-depleted cells (+Tet), no anti-VSG signals were observed in the enlarged flagellar pockets (t=1, t=2) (arrows, lower panels). There was considerable loss of anti-VSG signal between the t=0 and t=2 timepoints, and half of the TbSmee1-depleted cells displaying a phenotype had no observable signal at all, making the conclusions from these experiments somewhat tentative.

To quantify the results, the incidence of morphologically abnormal cells was counted, and then the various distributions of anti-VSG antibodies were classified into three categories (Fig. 8C). Of the 35% of cells in the TbSmee1-depleted population that displayed a phenotype after 24 h of RNAi, 26% were classified as being completely rounded up (‘BigEye’), whereas 7% had the less-developed ‘enlarged pocket’ state. The anti-VSG antibody distributions were grouped into three categories: a dot signal, a different signal (usually a dot with other weaker labelling) or no signal, respectively. Although most of the morphologically abnormal cells had no anti-VSG signal at all, TbSmee1-depleted cells with clearly internalised anti-VSG of the kind observed in control cells were never seen.

In summary, the inhibition of endocytosis caused by depletion of TbSmee1 prevents the internalisation of surface-bound anti-VSG antibodies and appears to also prevent their entry into the enlarged flagellar pocket.

In this study, the hook complex component TbSmee1 was characterised in bloodstream form T. brucei. It is only the third hook complex protein to be characterised in this life cycle stage after TbMORN1 and Bhalin and, similar to them, it is essential for the viability of the cells in vitro (Broster Reix et al., 2021; Morriswood and Schmidt, 2015). The localisation of TbSmee1 was found to be identical to that previously documented in the procyclic life cycle stage of T. brucei (Figs S3, S2) (Perry et al., 2018). It is constitutively present at the shank part of the hook complex and additionally localises to the tip of the growing new FAZ in replicating cells. At the time of the preprint of this work being published, TbSmee1 was the only protein other than the DOT1 antigen and FLAM3 to be shown to localise to the groove structure associated with the tip of the growing new FAZ (Sunter et al., 2015a). A number of additional proteins present at the groove have since been identified (Smithson et al., 2022). It is unclear what role the phosphorylation of TbSmee1 plays in its localisation, but, given that it is both a substrate and putative direct binding partner of the mitotic kinase TbPLK, it is possible that its phosphorylation status might be related to its localisation to the groove structure.

Phenotypic characterisation of TbSmee1 showed that its depletion results in the enlargement of the flagellar pocket with concomitant effects on the entry of macromolecular cargo (Figs 4–6). There was extremely good agreement between the immunoblotting, growth curve and cell cycle analysis data. At 18 h post induction of RNAi, TbSmee1 protein levels had nearly reached their minimum but there was still no observable effect on cell cycle progression or population growth. From around 20–21 h post induction, when TbSmee1 levels reached 10–15% of those in controls, cell cycle progression slowed, resulting in an increase in the number of 2K1N and 2K2N cells. This was mirrored by a slowing of population growth. Systematic analysis of a number of marker proteins for the hook complex, the centrin arm and the flagellar pocket collar showed that none were strongly affected in terms of their protein levels or localisation by the depletion of TbSmee1. The lack of observable structural changes to the hook complex, flagellar pocket collar and centrin arm following TbSmee1 depletion implied that the lethal phenotype was due to a loss of a specific cellular function, rather than destabilisation of structural complexes.

In the previous characterisation of TbSmee1 in the procyclic form life cycle stage, depletion of TbSmee1 was found to cause structural changes to the hook complex and altered TbMORN1 distribution patterns (Perry et al., 2018). These analyses were, however, done after 6 days of RNAi induction, when TbSmee1 levels were already undetectable from day 1 (a growth defect was observable following 3 days of RNAi). It is very likely that similar changes would be observed in the bloodstream form stage at later RNAi timepoints, although the essential nature of TbSmee1 in bloodstream form cells might make this analysis difficult.

Flow cytometry analysis indicated that the flagellar pocket enlargement began around 18 h post induction, prior to the effects on population growth and cell cycle progression. This therefore suggests that flagellar pocket enlargement is an early and likely direct consequence of TbSmee1 depletion, with the effects on cell cycle progression coming afterwards. One plausible hypothesis to account for these observations is that the spatial problems produced by flagellar pocket enlargement impair cell cycle progression. The relative lack of cell cycle checkpoints means that ‘monster’ cells with re-duplicated organelles are then produced. The internal pressure caused by continuing enlargement of the flagellar pocket forces the cells into an increasingly spherical shape and, eventually, cell viability is lost.

The enlargement of the flagellar pocket was accompanied by a cargo access defect previously documented for TbMORN1 and Bhalin – small fluid-phase cargo such as 10 kDa dextran filled the enlarged flagellar pocket, whereas larger cargo such as ConA and BSA did not (Broster Reix et al., 2021; Morriswood and Schmidt, 2015). In TbSmee1-depleted cells, BSA was observed in one or more small foci next to the flagellar pocket. This was surprising, as BSA is known to traffic in the fluid phase and should not be binding to the cell surface. BSA was, however, also observed to associate with the flagellar pocket membrane in an electron microscopy-based study of cargo uptake (Gadelha et al., 2009). In a classic radiochemical study of cargo uptake in T. brucei, it was observed that “at low concentration, a small adsorptive component may become prevailing” in assays using BSA (Coppens et al., 1987). As the experiments conducted in this study, in Coppens et al. (1987) and in Gadelha et al. (2009) used BSA well below physiological concentrations, it seems likely that these foci represent the small adsorptive component that was indirectly observed in the biochemical assays. Therefore, in these assays, the ConA and BSA probes both report on the behaviour of surface-bound cargo, whereas the dextran reports on fluid-phase uptake.

As ConA is not a physiological cargo, the entry of anti-VSG antibodies into the flagellar pocket was also investigated. These assays turned out to be unexpectedly difficult, owing to much of the signal being lost during the wash steps. This might be related to the recently reported high rate of VSG shedding (Garrison et al., 2021). Nonetheless, the t=0 and t=2 timepoints appear to match expectations: in control cells, the anti-VSG antibodies were taken up by endocytosis, whereas in TbSmee1-depleted cells, any remaining signal was found outside the area of the enlarged flagellar pocket and apparently still on the cell surface.

When the TbMORN1 phenotype was characterised (Morriswood and Schmidt, 2015), this was the first time that a protein had been shown to play a role in the entry of surface-bound cargo into the flagellar pocket. Subsequent characterisation of Bhalin (Broster Reix et al., 2021) and the work here on TbSmee1 show that the cargo entry defect appears to be a consistent effect after enlargement of the flagellar pocket. That the same cargo entry defect is seen upon depletion of clathrin suggests that this is a previously unobserved feature of all flagellar pocket enlargement (‘BigEye’) phenotypes. Therefore, active endocytosis appears to be required for the entry of surface-bound cargo into the flagellar pocket.

It should be noted that the results obtained in the clathrin RNAi experiments contradict previously published observations, where ConA was claimed to accumulate inside the enlarged flagellar pocket of clathrin-depleted cells (Allen et al., 2003). Re-examination of previously published data from clathrin RNAi cells suggests, however, that the effect has always been present, with ConA signal (and indeed other cargoes) absent from the flagellar pocket interior and predominantly concentrated in one or more foci just outside it [see, for example, Fig. S4 in Allen et al. (2003) and Fig. S2 in Zoltner et al. (2015)]. This oversight is understandable, given that the focus in these clathrin papers was on whether cargo endocytosis was occurring, and not whether the cargo was able to access the flagellar pocket (Mark Field, University of Dundee, personal communication). Importantly, it also means that similar observations have been made by (at least) two independent groups.

This raises two important questions. First, why does inhibition of endocytosis apparently hinder the entry of surface-bound cargo into the flagellar pocket? Second, why does the depletion of hook complex proteins – which are spatially removed from the actual sites of clathrin-coated vesicle formation – cause an endocytosis defect? The prevailing model for several years has been that hydrodynamic flow, driven by flagellar motility, is responsible for sorting surface-bound cargo to the posterior end of the cell where the flagellar pocket is located (Engstler et al., 2007). Once there, the surface-bound material enters the flagellar pocket through a narrow channel where the flagellar membrane and the flagellar pocket neck membrane are not as closely apposed (Gadelha et al., 2009). The results here suggest that the posterior sorting mechanism still functions when endocytosis is inhibited, but that the transit of surface-bound material through the channel and into the flagellar pocket is impeded.

In this context, it is worth considering the somewhat confusing anti-VSG uptake results obtained at the t=1 timepoint in control cells, in which the anti-VSG signal came from a single spot adjacent to the kinetoplast. As a temperature block was in place to prevent endocytosis, this would at first sight suggest that the anti-VSG antibodies can enter the flagellar pocket in the absence of endocytosis and that the entire hypothesis proposed here is invalid. Given, however, that the flagellar pocket is less than 1 µm across, it is possible that there is insufficient resolution to distinguish the entrance to the flagellar pocket from the flagellar pocket itself in control or wild-type cells. In this interpretation, it is only when the flagellar pocket is expanded due to the BigEye phenotype that these two localisations can be resolved at normal widefield resolution. Higher-resolution imaging of these cells is therefore an important target for follow-up work.

An earlier hypothesis for hook complex function was that it somehow maintained the integrity of the channel in the flagellar pocket neck. Nevertheless, the fact that the same phenotype was obtained following depletion of clathrin suggests that the hook complex instead indirectly affects endocytosis. This could potentially be by affecting either the localisation or activity of a number of lipid kinases, at least two of which are known to localise to the hook complex (Dean et al., 2017; Demmel et al., 2014). The activity of these enzymes could then license the flagellar pocket membrane for endocytosis, for instance, by generating the essential endocytic cofactor phosphatidylinositol-(4,5)-bisphosphate. Subsequent internalisation of the membrane by endocytosis would then assist to pull in more membrane currently at the flagellar pocket entrance. How this would be integrated with the activity of the exocytic pathway is, however, unclear.

Thus, although hydrodynamic flow might be responsible for concentrating cargo at the entrance to the flagellar pocket, endocytic activity appears to be required for the entry of the cargo into the flagellar pocket itself, and endocytosis might assist or be responsible for pulling material in through the channel.

The TbSmee1(1–400) open reading frame was amplified from Trypanosoma brucei brucei strain Lister 427 genomic DNA by PCR. The PCR product was ligated into the p3NH expression vector [a derivative of the commercially available plasmid pACE1 (Geneva Biotech)], which encodes an N-terminal His6 tag, using sequence and ligation-independent cloning (Li and Elledge, 2012). The plasmid was used to transform Escherichia coli strain Rosetta II (DE3)pLysS by heat shock, and individual colonies were subsequently grown at 37°C in the presence of 100 μg/ml kanamycin to an optical density at 600 nm of ∼0.8–1.0. Recombinant protein expression was induced by the addition of 50 μM isopropylthio-β-galactoside, and the cells were then incubated overnight at 20°C with shaking. Cells were harvested by centrifugation (5000 g for 30 min). The pooled pellet from 6 l of cell culture was resuspended in 300 ml of lysis buffer [50 mM HEPES pH 7.0, 500 mM NaCl, 20 mM imidazole, 5% glycerol, 0.5% Triton-X, 1 mM tris(2-carboxyethyl)phosphine (TCEP), 200 mM phenylmethylsulfonyl fluoride and protease inhibitor cocktail]. Pellet emulsions were first homogenised by mixing on ice using a T 10 basic Ultra-Turrax dispersing instrument (IKA). Final lysis was then achieved with sonication on ice, using three cycles of 3 min at 50% strength. Lysates were clarified by centrifugation (18,000 g, 45 min, 4°C). The lysates were added to a HiTrap Chelating HP 5 ml column (GE Healthcare) equilibrated with buffer A (20 mM HEPES pH 7, 300 mM NaCl, 40 mM imidazole, 2% glycerol and 1 mM TCEP), and eluted with a 100% step gradient of buffer B1 (20 mM HEPES pH 7, 300 mM NaCl, 400 mM imidazole, 2% glycerol and 1 mM TCEP). Selected peak fractions were examined by SDS-PAGE for protein content and purity. Fractions containing a dominant band at approximately 46 kDa were pooled and concentrated using Amicon Ultra centrifugal filter units with 10K pore size (MerckMillipore) according to the manufacturer's instructions. The His6 tag was removed by 3C protease during overnight dialysis in dialysis buffer (20 mM Tris-HCl pH 7, 300 mM NaCl, 2% glycerol, 1 mM dithiothreitol). Significant losses were incurred during this step. TbSmee1(1–400) was applied to a previously equilibrated HiTrap Chelating HP 5 ml column charged with 50 mM CoCl₂ and coupled to a GSTrap HP 1 ml column (both GE Healthcare). Buffer A was used for equilibration. TbSmee1(1–400) was mostly collected from the flowthrough and a few initial collected fractions. Selected peak fractions were examined by 15% SDS-PAGE for protein content and purity. Fractions containing a dominant band at approximately 44 kDa were pooled and concentrated in Amicon Ultra centrifugal filter units (10K pore size) according to the manufacturer's instructions. Finally, TbSmee1(1–400) concentrates were applied to a previously equilibrated HiLoad 16/600 Superdex 200 pg column (GE Healthcare) pre-equilibrated in dialysis buffer. Flow speed was adjusted to 1 ml/min and fractions of 1.5 ml were collected. Fractions corresponding to the targeted chromatographic peak were examined for protein content by 15% SDS-PAGE, pooled accordingly to their purity, concentrated and stored at −80°C.

Purified recombinant TbSmee1(1–400) was used for the generation of two polyclonal rabbit antisera (Eurogentec). Antisera (303 and 304) were initially affinity purified against the TbSmee1(1–400) antigen, but the neat antisera were later found to show high specificity and were also occasionally used. A third polyclonal antibody (508) was generated against two TbSmee1 peptides (Eurogentec) and affinity purified using the peptide antigens immobilised on a Sulfolink affinity column (Thermo Fisher Scientific). Results shown were predominantly obtained using affinity-purified anti-TbSmee1 antibodies 303 and 304; most immunoblotting data were generated using the affinity-purified antibody 304 (1:200), as this had the lowest background in this application, whereas labelling was identical for antibodies 303 (1:100) and 304 (1:200) in immunofluorescence experiments. The results obtained with all three antibodies were consistent. The anti-Starkey1 anti-peptide rabbit polyclonal antibodies were generated and affinity purified in the same way [used at 1:200 for immunoblotting (IB) and 1:2000 for immunofluorescence (IF)].

The following primary antibodies have been described previously: rabbit anti-TbMORN1 (1:5000 for IF, 1:10,000 for IB; Morriswood et al., 2013), rabbit anti-TbBILBO1 (1:100 for IB; Esson et al., 2012), mouse anti-Ty1 (BB2; 1:100 for IB, 1:50 for IF; gifted by Cynthia He, National University of Singapore; Bastin et al., 1996), mouse anti-TbLRRP1 (1:500 for IF; gifted by Cynthia He; Zhou et al., 2010), mouse anti-TbCentrin4 (6C5; 1:50 for IB, 1:300 for IF; gifted by Chris de Graffenried, Brown University; Ikeda and de Graffenried, 2012), mouse anti-TbCentrin2 (2B2H1; 1:50 for IB, 1:100 for IF; gifted by Chris de Graffenried; de Graffenried et al., 2013), mouse anti-TbFAZ1 (L3B2; 1:50 for IF; gifted by Keith Gull, University of Oxford; Kohl et al., 1999), mouse anti-PFR1/2 (L13D6; 1:200 for IB; gifted by Keith Gull; Kohl et al., 1999), mouse anti-FAZ filament (DOT1; used undiluted for IF; gifted by Keith Gull; Woods et al., 1989), rabbit anti-VSG(221) (1:5000 for IF; gifted by Markus Engstler, University of Würzburg; Batram et al., 2014), anti-BiP (1:10,000 for IB; gifted by Jay Bangs, University at Buffalo) and anti-p67 (1:500 for IF; gifted by Jay Bangs). The following antibodies came from commercial sources: goat anti-rabbit(IRDye800CW) (926-32211; 1:10,000; LI-COR), goat anti-mouse(IRDye680LT) (926-68020; 1:10,000; LI-COR), and goat anti-rabbit and anti-mouse antibodies conjugated to Alexa Fluor dyes (A-11008, A-11001, A-11011, A-11004; 1:3000; Molecular Probes).

Wildtype Lister 427 (monomorphic) bloodstream form cells were cultured in HMI-9 medium (Hirumi and Hirumi, 1989) supplemented with 10% foetal bovine serum (FBS), 100 U/ml penicillin and 0.1 mg/ml streptomycin at 37°C and 5% CO₂. The single-marker (SM) cells (Wirtz et al., 1999) were cultured in the presence of G418 (2.5 μg/ml; VWR). Population density was monitored using a Z2 Coulter Counter (Beckman Coulter) and kept below 2×10⁶ cells/ml.

The Ty1–TbSmee1 endogenous replacement cell line was generated by transfection of 427 cells with a targeting fragment containing 285 bp of the TbSmee1 5′ untranslated region (UTR), a blasticidin resistance gene, the intergenic region of α- and β-tubulin, the sequence for a triple Ty1 tag and the first 399 bp of the TbSmee1 open reading frame without the start codon. Clones were selected by growth in medium containing 5 μg/ml blasticidin. RNAi target sequences were chosen using RNAit (Redmond et al., 2003). TbSmee1 RNAi cells were generated by cloning the RNAi target sequence into the pGL2084 plasmid (Jeremy Mottram, University of York) (Jones et al., 2014) and then transfecting 2T1 cells with the linearised plasmid (Alsford et al., 2005). Clones were selected by growth in medium containing phleomycin (2.5 μg/ml) and hygromycin (5 μg/ml). TbStarkey1 RNAi cells and clathrin (TbCHC) RNAi cells were generated by cloning the relevant RNAi target sequence into the p2T7_TAblue plasmid (David Horn, University of Dundee) (Alibu et al., 2005) and then transfecting SM cells with the linearised plasmid. Clones were selected by growth in medium containing G418 (2.5 μg/ml) and hygromycin (5 μg/ml). Ty1–TbSmee1 truncation constructs were cloned into the pLew100_v5-Hyg plasmid (George Cross, Rockefeller University) using in vivo assembly (Watson and Garcia-Nafria, 2019). SM cells were transfected with the linearised plasmid. Clones were selected by growth in medium containing G418 (2.5 μg/ml) and hygromycin (5 μg/ml). For transfection, >2.5×10⁷ cells of the parental strain were washed and resuspended in 100 μl transfection buffer (90 mM Na₂PO₄, 5 mM KCl, 0.15 mM CaCl₂, 50 mM HEPES-NaOH, pH 7.3) containing 10 μg DNA and transfected by electroporation using an AMAXA Nucleofector Device (Lonza) with the program ‘X-001 free choice’. Transfected cells were incubated in 50 ml HMI-9 medium without selection overnight. The next day, drug selection was applied and clones were selected by limiting dilution. At least three separate clones of all cell lines were isolated to control for biological variability. Integration of targeting fragments at endogenous loci or the presence of Ty1–TbSmee1 truncation constructs in the genome was confirmed by PCR analysis of genomic DNA. Genomic DNA was isolated using a DNeasy Blood & Tissue kit (QIAGEN), and relevant products were amplified by PCR. All cloned constructs used for cell line generation had their DNA sequence confirmed by sequencing.

For generation of dephosphorylated whole-cell lysates, cells were harvested by centrifugation (1000 g, 10 min) and the cell pellet was resuspended in 1 ml PBS with 46 mM sucrose and 10 mM glucose (vPBS) containing EDTA-free protease inhibitors (Roche). The washed cells were pelleted by centrifugation (750 g, 4 min). The cells were then resuspended in lysis buffer (0.5% IGEPAL, 0.1 M PIPES-NaOH pH 6.9, 2 mM EGTA, 1 mM MgCl₂, 0.1 mM EDTA and EDTA-free protease inhibitor cocktail) to a final concentration of 4×10⁵ cells/μl and incubated for 15 min at RT on an orbital mixer to allow dephosphorylation to occur. SDS-loading buffer was then added to a final concentration of 2×10⁵ cells/μl, the samples were further denatured by boiling (100°C, 10 min), and then stored at −20°C. SDS-PAGE was carried out using a Mini-Protean Tetra Cell (Bio-Rad) and protein transfer to nitrocellulose membranes using a Mini-Trans blot cell (Bio-Rad). Protein transfer and equal loading were confirmed using REVERT total protein stain (LI-COR) according to the manufacturer's instructions. Membranes were blocked using blocking buffer (PBS with 0.3% Tween 20 and 10% milk) (30 min, RT, rocker). The membranes were then incubated in primary antibodies diluted in blocking buffer (1 h, RT, roller). After three washes in immunoblotting buffer (PBS with 0.3% Tween 20) the membranes were incubated with IRDye-conjugated secondary antibodies diluted in immunoblot buffer (1 h, RT, roller). After another three washes in immunoblot buffer, the membranes were visualised using an Odyssey CLx (LI-COR). Background subtraction, qualitative analysis, normalisation relative to total protein stain and quantification were carried out using Image Studio Lite 5.2 and Empiria Studio 1.1 (LI-COR). Dot plots were generated using the web app PlotsOfData (Postma and Goedhart, 2019).

427 bloodstream form cells were grown to approximately 1.5×10⁶ cells/ml and harvested by centrifugation (1000 g, 10 min, 4°C). The cells were washed in 1 ml ice-cold wash buffer (0.1 M PIPES-NaOH pH 6.9, 2 mM EGTA, 1 mM MgCl₂, 0.1 mM EDTA and phosphatase inhibitor cocktail 2) and pelleted by centrifugation (750 g, 3 min, 4°C). For extraction of the cytoskeletons, the cell pellet was resuspended in 1 ml ice-cold extraction buffer (0.5% IGEPAL, 0.1 M PIPES-NaOH pH 6.9, 2 mM EGTA, 1 mM MgCl₂, 0.1 mM EDTA, EDTA-free protease inhibitor cocktail and phosphatase inhibitor cocktail 2) and incubated for 15 min on ice. The cell suspension was inverted every 5 min. The cytoskeletons were separated from the cytoplasm by centrifugation (750 g, 3 min, 4°C). The cytoskeletons were washed with 0.5 ml ice-cold 1× NEBuffer for PMP (New England Biolabs, P0753S) supplemented with 1 mM MnCl₂ and pelleted by centrifugation (750 g, 3 min, 4°C). After resuspension in 375 μl ice-cold 1× NEBuffer for PMP supplemented with 1 mM MnCl₂, all following samples were taken from this stock. An input sample (0 min) of 40 μl was taken and added to 20 μl of SDS-loading buffer. Two control samples were taken, 0.4 μl phosphatase inhibitor cocktail 2 was added, and the reactions were incubated on ice and at 26°C, respectively, for 20 min. 20 μl of SDS-loading buffer was then added. To assay for dephosphorylation, 130 μl was taken from the stock and 1 μl λ-phosphatase (400 U; New England Biolabs, P0753S) was added. This sample and the remainder from the stock were incubated at 26°C. After 1, 2, 3, 5, 10 and 20 min, 20 μl samples were taken from each reaction and were added to 10 μl SDS-loading buffer. All the samples were boiled at 104°C for 10 min and stored at −20°C. Samples were analysed by immunoblotting.

427 bloodstream form cells (50 ml) were grown to approximately 1.5×10⁶ cells/ml and then harvested by centrifugation (1000 g, 10 min, 4°C). The cell pellet was resuspended in 1 ml vPBS, transferred to a microfuge tube, and the cells pelleted again by centrifugation (750 g, 4 min, 4°C). The supernatant was discarded and the centrifugation step was repeated to remove remaining supernatant. The cell pellet was resuspended in 200 μl extraction buffer (0.5% IGEPAL, 0.1 M PIPES-NaOH pH 6.9, 2 mM EGTA, 1 mM MgCl₂, 0.1 mM EDTA and EDTA-free protease inhibitor cocktail) and incubated for 15 min at RT in an orbital mixer. A 5% input sample was taken (10 μl), put into a new microfuge tube and left on ice. The extracted cells were fractionated by centrifugation (3400 g, 2 min, 4°C) and the supernatant transferred to a fresh microfuge tube and the exact volume noted. The tube containing the extracted cells was centrifuged again at the same settings and this second residual supernatant discarded. The cytoskeleton pellet was then resuspended in 200 μl extraction buffer. 5% samples of supernatant and pellet fractions were taken and analysed by immunoblotting. Equal fractions of I, SN, P were loaded in each lane (I∼1.4×10⁶ cells).

Coverslips were washed in 70% ethanol and then incubated with 0.01% poly-L-lysine in a 24-well plate (>20 min, RT) and left to dry. 2×10⁶ cells were taken per coverslip and transferred to 15 ml Falcon tubes. The cells were pelleted by centrifugation (1000 g, 1 min per ml of liquid, RT) in a swing-bucket centrifuge. The supernatant was removed, and the cell pellet was gently resuspended in 1 ml ice-cold vPBS+cOmplete EDTA-free protease inhibitor cocktail (Roche). The cells were again pelleted by centrifugation (1000 g, 2 min, RT) and subsequently resuspended in 1 ml ice-cold vPBS+cOmplete EDTA-free protease inhibitor cocktail and directly added to the coverslips. The cells were attached to the coverslips by centrifugation (1000 g, 1 min, RT), and attachment was confirmed visually. The attached cells were then incubated in 1 ml ice-cold extraction buffer (0.5% IGEPAL 0.1 M PIPES-NaOH pH 6.9, 2 mM EGTA, 1 mM MgCl2, 0.1 mM EDTA and EDTA-free protease inhibitor cocktail) (5 min, on ice). The extracted cells were washed two times with 1 ml vPBS with cOmplete EDTA-free protease inhibitor cocktail and then fixed in 1 ml ice-cold 99.9% methanol (30 min, −20°C). The fixed cells were rehydrated using 1 ml PBS. The coverslips were blocked in 1 ml 3% BSA in PBS (30 min, RT), and sequentially incubated with clarified primary and secondary antibodies diluted in PBS (1 h, RT, humidified chamber for each) with three PBS washing steps (5 min, RT, rocker) after each incubation. After the final wash, glass slides were cleaned with 70% ethanol, and a spot of DAPI Fluoromount-G (Southern Biotech) was placed on the surface. The coverslips were rinsed in ddH₂O, carefully dried and then mounted. For preparation of samples for SIM imaging, different coverslips (high precision, no. 1.5H) were used. Cells were harvested as above, washed, and fixed in 4% paraformaldehyde solution (10 min, ice then 30 min, RT). The fixed cells were washed, attached to coverslips, permeabilised (0.25% Triton X-100 in PBS; 5 min, RT), and then labelled and mounted as described above.

Images were acquired using a DMI6000B widefield microscope (Leica Microsystems, Germany) with a HCX PL APO CS objective (100×, NA=1.4, Leica Microsystems) and Type F Immersion Oil (refractive index=1.518, Leica Microsystems). The microscope was controlled using LAS-X software (Leica). Samples were illuminated with an EL6000 light source (Leica) containing a mercury short-arc reflector lamp (HXP-R120W/45C VIS, OSRAM, Germany). Excitation light was selected by using Y3 (545/25 nm), GFP (470/40 nm) and A4 (360/40 nm) bandpass filter cubes (Leica Microsystems). The power density, measured at the objective focal plane with a thermal sensor head (S175C, Thorlabs), was, respectively, 0.749±0.086, 0.557±0.069 and 0.278±0.076 W/cm² for the three filters. Emitted light was collected at ranges of 605/70 (Y3), 525/50 nm (GFP) and 470/40 nm (DAPI), respectively. The individual exposure times and camera gains were adjusted according to the different samples. RNAi samples (control and depleted) were imaged using identical settings. Differential interference contrast (DIC) microscopy was used to visualise cell morphology. A three-dimensional image of each field of view was obtained using 40 z-slices (step size=0.21 µm). Fields of view were selected in the DIC channel in order to mask the user to the fluorescence signal and subjectively select for cells with optimum morphology. Images were captured using a DFC365 FX monochrome CCD camera (Leica, 6.45 µm pixel size). SIM images were acquired using an Elyra S.1 SIM microscope (Zeiss) and ZEN software (Zeiss).

Processing was carried out using FIJI (Schindelin et al., 2012) and a custom macro for the generation of maximum-intensity z-projections with single DIC z-slices overlaid. Overlaps between two proteins were confirmed in individual z-slices (thickness=210 nm). The plugin ScientiFig was used to create the collage and adding the inserts (Aigouy and Mirouse, 2013). For correlation analysis between TbSmee1 and other flagellar pocket collar and/or hook complex-associated proteins, 1K1N cells were selected using both DIC and DAPI channels. Two-dimensional (2D) sum-slice projections were prepared for each stack of both green and red channels. The projections were clipped to 8-bit depth and a convoluted background subtraction was applied. All resultant individual 2D images, without channel overlay, were analysed pairwise to check for intensity-based correlation based on Zhang and Cordelières (2016). The Spearman's rank correlation results were further computed into Microsoft Excel sheets and analysed using R version 4.1.2 (https://www.r-project.org/) in the environment RStudio 2021.09.1.372 (https://www.r-project.org/conferences/useR-2011/abstracts/180111-allairejj.pdf). The packages used for all descriptive analysis and plot generation were ggplot2 (Wickham, 2016), openxlsx (https://cran.r-project.org/web/packages/openxlsx/) and psych (https://cran.r-project.org/web/packages/psych/index.html). All FIJI/ImageJ macros and R scripts for correlation analysis were written by A.B. and are available upon request.

RNAi cells were seeded at the required starting concentration in a volume of 22 ml and divided into two 10 ml aliquots in separate flasks. Tetracycline was added to a final concentration of 1 μg/ml in one flask to induce RNAi, and refreshed every 24 h. The population density of the control and induced cells was measured every 24 h over a time course of 72 h, or every hour over a time course of 8 h, using a Z2 Coulter Counter (Beckman Coulter). Depletion of the target protein was confirmed in every experiment by immunoblotting of whole-cell lysates.

An aliquot of 10⁶ cells was taken, and the cells were fixed directly in medium by addition of isothermal 25% glutaraldehyde to a final concentration of 2.5% (10 min, 37°C, gentle mixing). The cells were then pelleted by centrifugation (750 g, 10 min). The cell pellet was resuspended in 0.5 ml 2.5% glutaraldehyde in PBS, transferred to a microfuge tube and incubated at RT (30 min, gentle mixing). The cells were pelleted again by centrifugation (750 g, 4 min) and the cell pellet was resuspended in 500 μl PBS. The cells were then added to the coverslips inside the 24-well plate and attached by centrifugation (1000 g, 1 min, RT). The coverslips were mounted on poly-L-lysine-coated slides using DAPI Fluoromount-G. Imaging was as described for fluorescence microscopy above, using DAPI and DIC channels only. Cell division cycle stages (1K1N, 2K1N and 2K2N) were manually quantified from maximum-intensity projections of the DAPI signal overlaid with single DIC z-slices. Depletion of the target protein at each timepoint was confirmed by immunoblotting of whole-cell lysates from the same experiments.

Induced and uninduced RNAi cells were grown for 24 h to a density of 1-2×10⁶ cells/ml in 50 ml medium and harvested by centrifugation (1000 g, 10 min, RT). All but 4 ml of the supernatant was removed and 4 ml FBS was added. The cells were pelleted again (1000 g, 10 min, RT) and all but 200 μl of the supernatant was removed. The cells were resuspended in the remaining supernatant, and the suspension was then transferred to a PCR tube and pelleted by centrifugation (1600 g, 10 s, RT). The cells were then transferred into a carrier with a closed lid to avoid air inclusions. High-pressure freezing was started immediately using a Leica EM HPM100. After high-pressure freezing, the samples were transferred to Leica EM AFS2 system for freeze substitution and progressive lowering of temperature. Low-temperature embedding and polymerisation of Epon resin {2-dodecenylsuccinic acid anhydride, methyl nadic anhydride, Epon 812 and 2,4,6 Tris[dimethylaminomethyl(phenol)]} were then performed. Ultra-thin cuts (60 nm) were performed with an ultramicrotome (Leica EM UC7/FC7) and sections were placed on slotted grids. For contrasting, sections were incubated in 2% uranyl acetate for 8 min. Afterwards, the grids were washed three times in ddH₂O (boiled to remove CO₂) and incubated for 5 min on 50% Reynold's lead citrate in a Petri dish with NaOH tablets. The grids were again washed two times in ddH₂O. A 200 kV transmission electron microscope (Jeol, JEM-2100) with a TemCam F416 4k×4k camera (Tietz Video and Imaging Processing Systems) and EMMenu 4.0.9.31 software (Tietz) were used. Uninduced control cells were viewed at a magnification of 12,000× and induced cells were viewed at a magnification of 8000×.

Induced and uninduced RNAi cells at a concentration of ∼5×10⁶ cells/ml were harvested by centrifugation (1000 g, 4°C). The cells were resuspended in 45 μl ice-cold vPBS containing protease inhibitors and incubated on ice to block endocytosis (10 min, ice). 5 µl labelled dextran (10 kDa, 50 mg/ml stock; Invitrogen, D22910) was added and mixed by flicking. The mixture was then incubated to allow dextran to enter and fill the flagellar pocket (15 min, on ice, dark). At the end of the incubation, 1 ml ice-cold vPBS with protease inhibitors was added and the cells were pelleted by centrifugation (750 g, 2 min, 4°C). The cell pellet was resuspended in 0.5 ml vPBS containing protease inhibitors, and the cells were fixed by addition of 0.5 ml 8% paraformaldehyde and 0.1% glutaraldehyde solution in vPBS containing protease inhibitors (20 min, on ice, then 60 min, RT). The fixed cells were pelleted by centrifugation (750 g, 2 min) and washed twice with 1 ml vPBS containing protease inhibitors. The dextran signal was immediately measured by flow cytometry using a FACSCalibur (BD Biosciences) running CellQuest Pro Software (BD Biosciences). Subsequent processing and analysis were carried out using FlowJo 10.8.1. Dot plots were created using Plots of Data (Postma and Goedhart, 2019). After flow cytometry, the remaining cells were attached to coverslips by centrifugation, mounted on clean glass slides using DAPI Fluoromount-G, and imaged using fluorescence microscopy on the same day to confirm flagellar pocket labelling. Depletion of TbSmee1 was confirmed in every experiment by immunoblotting of whole-cell lysates.

For cargo (ConA, dextran and BSA) uptake assays, 2×10⁶ cells per sample were harvested by centrifugation (1000 g, 10 min, 4°C) and washed in 1 ml ice-cold vPBS. The washed cells were pelleted by centrifugation (750 g, 2 min, 4°C) and resuspended in 100 µl ice-cold vPBS. The cells were then incubated at low temperature (20 min, ice) to block endocytosis. During this incubation, the labelled cargoes were prepared. The 50 mg/ml dextran aliquot (10 kDa Alexa Fluor 488 conjugate, Invitrogen, D22910) was mixed using a sonicator bath (10 min, 37 Hz) and then vortexed. The 0.5 mg/ml ConA aliquot (TMR conjugate, Molecular Probes) was clarified by centrifugation (11,000 g, 10 min, 4°C). Both cargoes were kept on ice and in the dark. At the end of the incubation, 10 µl of the 50 mg/ml dextran and 2 µl of the 0.5 mg/ml ConA were added to the chilled cells, mixed by flicking, and incubated at low temperature (15 min, on ice, in the dark). T=0 samples were quenched and fixed at this point; t=30 samples had an additional incubation to allow endocytosis of cargo (30 min, 37°C, in the dark) before quenching and fixing. To quench samples, 1 ml ice-cold vPBS was added, and the cells were pelleted by centrifugation (750 g, 2 min, 4°C). The pelleted cells were resuspended in 50 µl vPBS by flicking, and then fixed by addition of 0.5 ml ice-cold fix solution (4% paraformaldehyde, 0.1% glutaraldehyde in vPBS) (20 min, on ice, in the dark, then 30 min, RT, in the dark). The fixed cells were pelleted by centrifugation (750 g, 2 min, RT), washed in 2 ml vPBS, resuspended in 1 ml vPBS, and then attached to poly-L-lysine-coated coverslips by centrifugation. Coverslips were mounted on glass slides using DAPI Fluoromount-G and imaged immediately. Assays using BSA (Alexa Fluor 555 conjugate, Molecular Probes) were performed in the same way. The 20 mg/ml BSA aliquot was clarified by centrifugation (750 g, 1 min, RT) and 3 µl was added to the cells/dextran mixture. Depletion of TbSmee1 was confirmed by immunoblotting of whole-cell lysates in every experiment.

For the anti-VSG uptake assay, 2×10⁶ cells per sample were harvested by centrifugation (1000 g, 1 min per ml of liquid+1 min, 4°C) and washed in 1 ml ice-cold vPBS. The washed cells were pelleted by centrifugation (1000 g, 3 min, 4°C), resuspended in 50 µl ice-cold vPBS, and incubated at low temperature to block endocytosis (20 min, ice). During incubation, the primary and secondary antibodies were clarified by centrifugation (10,000 g, 10 min, 4°C). The primary antibody (anti-VSG antiserum) was diluted 1:50 in ice-cold vPBS. The secondary antibody was diluted to the required concentration (1:3000) in 3% BSA in PBS. Both were kept on ice in the dark until used. At the end of the incubation, 50 µl of diluted anti-VSG was added to the cells and quickly mixed by flicking, followed by an incubation to allow binding (15 min, ice). To fix cells for t=0, 0.5 ml ice-cold vPBS and 0.5 ml ice-cold 8% paraformaldehyde in vPBS were sequentially added at the end of the incubation and kept on ice. For timepoints t=1 and t=2, 1 ml ice-cold vPBS was added. The cells were washed by centrifugation (1000 g, 2 min, 4°C) and resuspended in 1 ml ice-cold vPBS. To fix cells for t=1, 350 µl of ice-cold 16% paraformaldehyde was added and the cells were kept on ice. The cells for t=2 were incubated (2 min, 37°C) to allow redistribution, flagellar pocket entry and endocytosis of the anti-VSG. 350 µl of ice-cold 16% paraformaldehyde was then added to fix the cells. Samples from all three timepoints were incubated on ice (20 min) after addition of paraformaldehyde, followed by a subsequent incubation at RT (30 min). The fixed cells were pelleted by centrifugation (1000 g, 2 min, RT), resuspended in 1 ml RT vPBS, and attached to poly-L-lysine-coated coverslips by centrifugation (1000 g, 1 min, RT). The attached cells were then permeabilised in 1 ml 0.25% Triton X-100 in PBS (5 min, RT). The permeabilised cells were washed two times in 1 ml PBS and the coverslips with cells were placed on a drop of clarified and diluted secondary antibody (70 µl) in a humidified chamber and incubated (1 h, RT, dark). The coverslips were washed three times in 1 ml PBS (5 min, RT, rocker, dark). After rinsing the cells in ddH₂O, they were mounted onto a drop of DAPI Fluoromount-G on a coverslip that was cleaned with 70% ethanol.

Antibodies were obtained from the following sources: Cynthia He (National University of Singapore) provided the anti-Ty1 (BB2) and anti-TbLRRP1; Chris de Graffenried (Brown University) provided anti-TbCentrin2 (2B2H1) and anti-TbCentrin4 (6C5); Markus Engstler (University of Würzburg) provided anti-VSG; Jay Bangs (University at Buffalo) provided anti-BiP and anti-p67; and Keith Gull (University of Oxford) provided anti-FAZ1 (L3B2), anti-PFR1/2 (L13D6) and DOT1. Nadine Weisert provided assistance with cloning. Philip Kollmannsberger assisted the correlation analysis. Balázs Szöőr (University of Edinburgh) and Mick Urbaniak (Lancaster University) provided input on the phosphatase assays. Mark Field (University of Dundee) provided extremely helpful advice and context on earlier clathrin RNAi work. Jay Bangs (University at Buffalo) provided helpful advice on the interpretation of the anti-VSG uptake data. Tim Krüger provided input and assistance with imaging. Helena Jambor (Technische Universität Dresden) advised on data visualisation. Graham Warren (Max Perutz Laboratories) hosted the preliminary experiments for this project. Susanne Kramer and Manfred Alsheimer provided feedback on the initial manuscript. Various members of the trypanosome research community provided feedback during the revision process. The transmission electron microscope was funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) – 426173797 (INST 93/1003-1 FUGG).

The peer review history is available online at https://journals.biologists.com/jcs/lookup/doi/10.1242/jcs.261548.reviewer-comments.pdf