TbKINX1B: a novel BILBO1 partner and an essential protein in bloodstream form Trypanosoma brucei

The flagellar pocket (FP) of the pathogen Trypanosoma brucei is an important single copy structure that is formed by the invagination of the pellicular membrane. It is the unique site of endo- and exocytosis and is required for parasite pathogenicity. The FP consists of distinct structural sub-domains with the least explored being the flagellar pocket collar (FPC). TbBILBO1 is the first-described FPC protein of Trypanosoma brucei. It is essential for parasite survival, FP and FPC biogenesis. In this work, we characterize TbKINX1B, a novel TbBILBO1 partner. We demonstrate that TbKINX1B is located on the basal bodies, the microtubule quartet (a set of four microtubules) and the FPC in T. brucei. Down-regulation of TbKINX1B by RNA interference in bloodstream forms is lethal, inducing an overall disturbance in the endomembrane network. In procyclic forms, the RNAi knockdown of TbKINX1B leads to a minor phenotype with a small number of cells displaying epimastigote-like morphologies, with a misplaced kinetoplast. Our results characterize TbKINX1B as the first putative kinesin to be localized both at the basal bodies and the FPC with a potential role in transporting cargo along with the microtubule quartet.


Introduction
Trypanosoma brucei is a zoonotic pathogen and the etiological agent of sleeping sickness with 60 million people living in areas with a high risk of infection, even though fewer than 2000 cases are detected per year. The procyclic form (PCF) and bloodstream form (BSF) T. brucei cell has a polarized organization and importantly, within its posterior end, it houses a unique structure called the flagellar pocket (FP) from which the flagellum emerges. This organelle-like structure is the exclusive site for endo-and exocytotic activity [32].
Within eukaryotic cells, transport of material often involves the activity of microtubule-dependent processes mediated by molecular motors such as kinesins (KIN) or dyneins. While most kinesins transport cargo through their interaction with microtubules (MT) by generating force through ATP hydrolysis [24,25], some of these motor proteins depolymerize MTs and participate in regulating MT dynamics [33]. A comprehensive phylogenetic analysis performed in a wide range of eukaryotes has allowed a revised classification of kinesins. Currently, there are 17 kinesin families (kinesin-1 to -17), as well as 14 additional paralog groups that cannot be assigned to any of the kinesin families (kinesinX1 -14). Little information is available for the paralog groups, as they contain a few members [45]. Trypanosomes possess a large number of kinesin encoding genes (37 genes in T. brucei plus 9 highly divergent genes versus 8 plus 1 highly divergent in Plasmodium falciparum [44] and 30 in Homo sapiens), indicating that some of these proteins could play essential and specific roles in these parasites or be redundant [45]. Functional characterization of kinesins in T. brucei has shown multiple, distinct and essential roles in flagellar construction, cell morphology, organelle segregation and mitosis [20,27,12,13,45].
We are interested in the organisation of the flagellar pocket and associated structures which ensure tight cellular and molecular links between the endomembrane components, in particular the flagellar pocket collar (FPC) and the hook complex [34,36]. We previously identified and characterized the first FPC member, BILBO1 [9,21,41,42] and demonstrated that downregulation of BILBO1 by RNA interference (RNAi) prevents the formation of a new FP and FPC, and is lethal in both PCF and BSF cells [9]. To identify other FPC components, we used the structural information for BILBO1 to design a yeast-two hybrid (Y2H) genomic screen, using BILBO1 as bait. Using this protocol, we identified several partner proteins and among these is the kinetoplastid-specific putative kinesin Tb927.7.3000 (initially named FPC5 [21]). Tb927.7.3000 protein belongs to the same kinetoplastid-specific KINX1 clade [45] as TbKINX1, a flagella connector protein (also named FCP2) [40]. Tb927.7.3000 contains a typical kinesin sequence motif and is now referred to as TbKINX1B, according to the kinesin nomenclature [45]. The Y2H screen identified amino acids (aa) 517-715 as the BILBO1-binding domain (B1BD). We further showed that this interaction with BILBO1 was dependent on the two BILBO1 EF-hand calcium-binding motifs [21].
TbKINX1B is part of the cytoskeleton of the trypanosome and is localized at the basal bodies as well as the FPC area, similar to BILBO1. RNAi down-regulation of TbKINX1B expression shows little effect in PCF but is lethal in BSF, where abnormal cells with an enlarged flagellar pocket are seen. Intriguingly, the absence of TbKINX1B does not affect the localization of TbBILBO1 in both forms.
We propose that TbKINX1B could be involved in the transport of cargos from the basal bodies to the FPC along the MTQ.

Ethics
The use of animals for the generation of monoclonal antibodies was in accordance with the rules of the ethical committee of the University of Bayreuth and licensed by the Government of Lower Franconia (licence RUF-55.2.2-2532-2-12-46-13).

Cell culture
The PCF T. brucei 427 29.13 and BSF T. brucei 427 90.13 cell lines (named wild-type WT used as controls) both co-expressing the T7 RNA polymerase and tetracycline repressor were cultivated with the appropriate antibiotics and transfected as specified in [21]. The TbBILBO1 RNAi cell line was previously described in [9]. All tetracycline inductions were carried at 10 lg/mL.

Purification of TbKINX1B motor domain proteins
Wild type motor domain (MD) of TbKINX1B (TbKINX1B MD ) and mutant TbKINX1B-DP-loop MD , were cloned into pET32c vector for expression in frame with the coding sequence of N-terminal thioredoxin-6 histidine in E. coli BL21(DE3) using primers: 5 0 -tgtgtcgacaaatgacgtctcaaacgtcg-3 0 /5 0 -tgtgcggccgctcagcgctggtcttcgttgac-3 0 . Deletion of the P-loop domain (TbKINX1B-DP-loop MD ) was achieved by direct mutagenesis (QuickChange kit, Agilent), following the manufacturer's recommendations and with the use of the following primers: 5 0 -tcatgtttgtttgcgtactacagcatgattgggccc-3 0 /5 0 -gggcccaatcatgctgtagtacgcaaacaaacatga-3 0 . Protein expression was induced for 1 h at 37°C with 1 mM isopropyl-bthiogalactopyranoside (IPTG). Cells were harvested by centrifugation at 4000 Âg for 20 min and the pellet resuspended in binding buffer (20 mM Tris-HCl pH 7.4, 150 mM NaCl, 5% Glycerol, 5 mM imidazole supplemented with 1 mM PMSF (phenylmethylsulfonyl fluoride) and Protease inhibitor cocktail set III-EDTA-free (Calbiochem), and 200U Benzonase) and lysozyme 0.1 mg/mL). The mix was left 30 min on ice. Cells were lysed by sonication and the lysate was centrifuged at 10,000 Âg for 20 min at 4°C. Soluble recombinant proteins were loaded onto a His FF HiTrap column (GE Healthcare) and washed in binding buffer supplemented with 20 mM imidazole. Finally, the proteins were eluted with a 35-300 mM imidazole gradient in binding buffer. The same protocol was applied for the purification of the N-terminal thioredoxin-6 histidine tag (Trx-6His), which was used as a control in the Kinesin ATPase activity assay. Three independent purifications of each recombinant protein were performed and each was used for the Kinesin ATPase activity.

Production of anti-TbKINX1B mouse monoclonal
The C 0 -terminus of TbKINX1B (aa 821-1342) was expressed in E. coli XL1Blue bacteria, using the pTrcHis vector (Invitrogen). This truncation contained an N-terminal His-tag. The protein was purified under native conditions on NI-NTA columns (Qiagen) and used to inject three BALB/c mice. Mice were inoculated with the purified protein as follows: first injection: 50 lg in PBS mixed with complete Freund's adjuvant, intraperitoneally (IP). Second and third injections: 25 lg mixed with incomplete Freund's Adjuvant (IP). Fourth injection, 25 lg in PBS, (IP). Injections were done at intervals of three weeks and the mouse with best serum response by ELISA assays was used for myeloma fusion. Fusion of spleen and myeloma cells (P3X63-Ag8.653), was PEG-induced. Screening of hybridoma culture supernatants was done initially by ELISA assay, and positive clones were then screened by Western blot and immunofluorescence. The resulting antibody is an IgM (kappa light chain). Supernatants were collected and precipitated with 50% ammonium sulphate.

Kinesin ATPase activity
The ATPase activity of purified proteins (TbKINX1B MD , TbKINX1BDP-loop MD and TbKINX1BTrx-6His) was evaluated using the commercially available Enzyme Linked Inorganic Phosphate Assay (ELIPA, Cytoskeleton). The assay was done with the GST recombinant human kinesin heavy chain motor domain (KHC MD ) as a positive control in presence of MTs (+MTs), and as a negative control in absence of MT (ÀMTs) (Cat. #KR01, Cytoskeleton Inc.), and in the presence of taxol-stabilized microtubules (MT002, Cytoskeleton Inc.). Each condition (microtubules (MT) alone, kinesin alone and microtubules + kinesin) was performed in triplicate. Kinetic read-out was initiated upon the addition of ATP and followed using an OPTIMA plate reader at a fixed wavelength of 360 nm of absorbance. Proteins (including positive control) were used at 0.4 nM. The rate of ATPase, as nM per minute per mg of protein was calculated using a standard Pi curve, according to the manufacturer's recommendations. Independent experiments from independent protein batches purifications (n = 3) were used and the results were plotted using Prism Software, Version 5.
Bloodstream form TbKINX1B RNA interference (TbKINX1B RNAi ) cell line Bloodstream form cells Tb427.90.13 were transfected by electroporation with 10 lg of NotI linearized plasmid and selected with 2.5 lg mL À1 phleomycin 24 h post-transfection, followed by serial dilution for clonal selection.

Electron microscopy
Log-phase BSF control and TbKINX1B RNAi 48 h induced cells were fixed by adding fixatives directly to medium to a final concentration of 2.5% glutaraldehyde for 10 min, cells were collected and processed as in [37].

Immuno-electron microscopy
Log-phase T. brucei PCF cells were harvested by centrifugation 800 Âg for 10 min (min) and resuspended in 500 lL PBS on a clean sheet of Nescofilm. The cells were then adsorbed onto glow-discharged Formvar and carbon-coated grids for 15 min room temperature (RT). To prepare extracted flagella, cells on grids were first detergent-extracted in PEME buffer (100 mM PIPES, 1 mM MgSO4, 0.1 mM EDTA, 2 mM EGTA, pH6.9) plus 1% NP-40 (15 min, RT), rinsed with PEME, then further extracted in PEME buffer, 1% NP-40, 1 M KCl (20 min, 4°C). Grids were then rinsed four times (5 min, RT) in PEME buffer and fixed in 3% PFA in PEME buffer (5 min, RT) then rinsed four times (5 min, RT) in 100 mM glycine in PBS. After extraction and rinsing with glycine buffer as described above, grids were moved through 2 Â 10 min drops of blocking buffer (PBS with 1% fish skin gelatin and 0.1% Tween 20). Grids were then incubated with 25 lL of mAb (TbKINX1B diluted 1:50 and, if indicated in the figure, combined with rabbit polyclonal TbBILBO1 1:2000 [1]), in blocking buffer for 2 h at RT. After primary incubation, grids were blocked 4 times, 5 min each, as described above. Then, grids were incubated with goat anti-mouse IgM 15 nm gold conjugated secondary antibody diluted 1:50 (British Biotech, GAMM15) and if TbBILBO1 was used with protein A/G mix 6 nm gold conjugated (Electron Microscopy Sciences). Grids were then transferred on a drop of blocking solution 4 times 5 min in PBS, then fixed in 2.5% glutaraldehyde and negatively stained in 10 lL of 5% NanoVan (Methylamine Vanadate-Nanoprobes). Samples were viewed on a MET FEI TECNAI 12 TEM electron microscope.

Immunoprecipitation
Trypanosoma brucei 427 29.13 PCF cells were grown up to log phase. The input material for IP (8 Â 10 8 cells) was washed in ice-cold phosphate buffer saline (PBS) and resuspended in 2 mL of lysis buffer (150 mM NaCl, 1 mM DTT, 1% NP-40 (Igepal CA-630), 25 mM Tris-HCl, pH 7.6) complemented with complete protease inhibitors (Roche) and 1 mM PMSF. The lysate was left on ice for 15 min and then sonicated (Bioruptor Plus) for 5 cycles of 30 s, level 5 amplitude. The lysate was cleared by centrifugation (30 min at 5000 Â g) and soluble material was used for the IP. A total of 50 lL of Dynabeads-Protein G (Life Technologies) were cross-linked to TbKINX1B rabbit polyclonal antibody (Eurogentec) or an unspecific antibody rabbit a-myc polyclonal antibody, using dimethyl 3,3 0 -dithiobispropionimidate (DTBP, Thermo Scientific), according to the manufacturer's instructions. Soluble material was incubated with the beads overnight at 4°C, in batch. Unbound material was kept for protein analysis (Flowthrough, FT fraction), as well as the washes performed with the lysis buffer (W fraction). Beads were resuspended in 2Â Laemmli buffer for 20 min at room temperature and then boiled for 10 min. Analysis of each of the fractions was carried by SDS-PAGE and immunoblot using mouse anti-TbKINX1B and rabbit anti-BILBO1 antibodies.

Results
TbKINX1B is a putative N-kinesin with ATPase activity The gene Tb927.7.3000 [6] was previously identified in a genomic yeast two-hybrid library screen with TbBILBO1 as bait (Hybrigenics) [21] and codes for a putative-kinesin (named here TbKINX1B) that belongs to the paralog group, Kinesin-X1 of kinesins, as classified by Wickstead et al. [45].
TbKINX1B is a 1342 amino acids protein with a predicted molecular mass of 151.3 kDa. In silico characterization of the primary and secondary structure of TbKINX1B (Fig. 1A) identified an N-terminal motor domain (aa 4-496) with the predicted four motifs involved in nucleotide-binding: the P-loop (GxxxxGKT/S) for phosphate-binding loop, the N2 or Switch-I (NxxSSRS), the N3 or Switch-II (DLAGxE), and N4 (RvRP) motif [39] (Fig. 1A). The central region of TbKINX1B contains a coiled-coil domain (aa 501-862) including a leucine zipper motif (aa 756-784) and the aa 517-715 domain, identified in the Y2H screen as the TbBILBO1-binding domain (B1BD) [21]. Finally, a coiled-coil region (aa 952-1304) is present in the C-terminal part of the protein. Tertiary structural organization of TbKINX1B was obtained by submitting the full-length protein sequence to Phyre2 [29] for an alignment-based homology model. Results showed that full-length TbKINX1B shares 48% structural identity with Mus musculus KIF1A (PDB 1I5S, N-terminal kinesin) with 100% confidence in the proposed motor domain model (Fig. 1B).
In order to test TbKINX1B ATP hydrolysis activity in vitro when binding to MTs, we affinity purified the recombinant TbKINX1B motor domain (aa 4-496, Trx-6His TbKINX1B MD ) and the motor domain deleted of its P-loop motif ( Trx-6His TbKINX1B-DP-loop MD) . The proteins were expressed with a thioredoxin N-terminal tag and a 6-histidine tag ( Trx-6His TbKINX1B MD , Trx-6His TbKINX1B-DP-loop MD ) in Escherichia coli and affinity purified (Fig. S1A). The TbKINX1B MD protein hydrolyses ATP with a Vmax of 288 nMol of ATP hydrolyzed per minute per mg of kinesin (nM Â min À1 Â mg À1 ). Unfortunately, the purified motor domain was prone to degradation (Fig. S1B), which could explain the reduced activity measured. As expected, no ATPase activity was detected for the motor domain deleted of the P-loop motif ( Trx-6His TbKINX1B-DP-loop MD ). Taken together, these results indicate that TbKINX1B displays the hallmarks of a typical N-kinesin.
TbKINX1B localises to the basal bodies, the MTQ and the FPC Many, but not all, kinesins move along microtubules and are therefore not strongly attached to the microtubule cytoskeleton. Western-blot analysis on whole cells (WC), detergentextracted cytoskeleton (CSK) fraction and soluble fraction (S) of PCF and BSF using an anti-TbKINX1B, showed that TbKINX1B is expressed in both PCF and BSF T. brucei and almost quantitively associated with the cytoskeleton fraction, similar to TbBILBO1 (Fig. 1C). As TbKINX1B was identified as a binding partner of TbBILBO1 [21], we compared its subcellular localization to TbBILBO1 in PCF and BSF (Figs. 2A, 2B). We focussed on detergent extracted cytoskeletons which allow better visualization of the FPC and the basal body (BB) region. We localized TbKINX1B and TbBILBO1 using an indirect immunofluorescence assay (IFA) (Figs. 2A, 2B), with a combination of anti-TbKINX1B and anti-TbBILBO1 antibodies. In some cells, TbKINX1B was seen only in the BB region, whereas in other cells, a dual localization (BB and FPC) could be observed (asterisks). We quantified this localization pattern at different cell cycle stages (1 kinetoplast-1 nucleus 1K1N, 2 kinetoplasts-1 nucleus 2K1N, and 2 kinetoplasts-2 nuclei 2K2N) (Fig. 2C, 200 cells counted, n = 3). Our results showed that in 1K1N PCF and BSF cells, respectively, TbKINX1B is seen in the BB region only (blue bars) in 84% and 87% of the cells, and in the FPC and BB region (FPC + BB) (orange bars) in 16% and 13% of the cells. When cells progress through the cell cycle, the proportion of co-localization of TbKINX1B to FPC + BB doubles in cells that have duplicated their kinetoplast (42% and 39% in 2K1N cells and 32% and 36% in 2K2N cells). The basal body localization of TbKINX1B was further characterized using the marker FTZC, a transition zone protein [10] associated with the mature and pro-basal bodies (mBB and pBB), and the BB marker YL1/2 [30] (Fig. 2D). IFA was done on detergent extracted cytoskeletons and isolated flagella and showed good co-localization for TbKINX1B and YL1/2. We next determined TbKINX1B localization by immuno-electron microscopy analysis of isolated flagella and confirmed the localization on the BBs and the MTQ/FPC, where it co-localizes with TbBILBO1 (Fig. 2E). Taken together, these results show that TbKINX1B localizes at the BBs, the MTQ, and the FPC and that the localization of the pool of TbKINX1B appears to depend on the cell cycle stage, in both PCF and BSF cells.
TbKINX1B and TbBILBO1 interact in vivo and B1BD is sufficient to target the FPC To test the TbKINX1B-TbBILBO1 interaction in vivo, we immunoprecipitated TbKINX1B from PCF cell extracts using a rabbit polyclonal antibody raised against the full-length TbKINX1B (Fig. 3A). Western blotting analysis showed that TbSAXO, an axonemal protein [17], is detected in the input samples (I) and the flow through (FT), but not in the elution sample (E). In contrast, TbKINX1B and TbBILBO1 are both detected in the input samples (I) and the elution sample (E) of the immunoprecipitation assay using anti-TbKINX1B, but not in the mock elution of the assay using a non-specific control anti-myc antibody. TbBILBO1 is thus specifically co-immunoprecipitated with TbKINX1B, suggesting that they belong to the same protein complex or interact directly. The latter possibility is strongly supported by the Y2H data that demonstrated the TbBILBO1-TbKINX1B B1BD interaction [21].
We next generated PCF cell lines that are inducible for the ectopic expression of a C-terminal myc-tagged version of fulllength TbKINX1B (TbKINX1B myc , 156 KDa), the TbKINX1B motor domain (TbKINX1B myc MD , 62 KDa) or the TbKINX1B B1BD (TbKINX1B myc B1BD , 29 KDa) (Fig. 3). The expression of the constructs induced no growth defect over 3 days of induction (Fig. 3B) and did not affect the TbBILBO1 protein levels (Fig. 3C). TbKINX1B myc was immuno-detected in the BB/FPC area, as previously observed for WT TbKINX1B (Fig. 3D). The motor domain localized to the MT cytoskeleton  in situ as shown by co-labelling with an anti-tubulin antibody, supporting the MT-binding properties evidenced in the kinesin assay. In contrast, TbKINX1B myc B1BD exclusively co-localized with TbBILBO1. This strongly suggests that the TbKINX1B BILBO1-binding domain is sufficient to target the protein to the FPC and to TbBILBO1. The expression levels of full-length TbKINX1B myc tend to increase over time, whereas TbKINX1B myc MD or TbKINX1B myc B1BD protein levels appear to be reduced after 48 h and 72 h of induction. It is unclear why this is the case, but the expression of a truncated protein could be toxic or have cryptic and unwanted functions leading to down-regulation of expression or protein degradation.

TbKINX1B is involved in basal body positioning in procyclic cells
To elucidate the function of TbKINX1B in T. brucei, we characterized the effect of its down-regulation in PCF using tetracycline-inducible RNA interference (RNAi) [3] (Fig. 4). Immunoblotting confirmed that TbKINX1B protein levels were reduced after 24 h of induction and undetectable after a longer induction time (Fig. 4A, left panel). After RNAi induction, a significant reduction of the growth rate was observed (Fig. 4A, right panel).
Light microscopy observation of PCF WT and TbKINX1B RNAi cells after 24 h to 72 h knockdown revealed a small subset of the population with an abnormal number of kinetoplasts (K) or nuclei (N) and FPC, and/or with detached and multiple flagella (Fig. S2A). In these cells, the absence of FAZ detection at the new detached flagella suggests a FAZ assembly defect leading to flagella detachment from the cell body. The newly detached flagellum was often found located close to the old mother flagellum, suggesting a basal body segregation defect (Fig. S2A). All these abnormal cells represent less than 10% of the population and are unlikely to have a significant effect on population growth within the observed period. TbKINX1B RNAi cells, after 24 h RNAi induction, display an abnormal positioning of the kinetoplast phenotype as shown by DAPI staining (Fig. 4B, graph). Quantification of kinetoplast positioning, in 1K1N cells, with reference to the nuclei (N) indicated that after 24 h induction, the spatial organization of the K within the population had changed and consisted of three groups (Fig. 4B). The first was "normal" (41.7%) whereby the K was located in the posterior of the cell as in WT 1K1N cells. The second was "close" (29.5%), in which the K was at the posterior of the cell but close to the nucleus. The third was "misplaced" (28.7%), where the K was located close to the nucleus but had relocated to the periphery of the cell in between the nucleus and the sub-pellicular MTs or anterior to the nucleus. The percentage of these subsets of cells continued to increase over the period of 48 h to 72 h RNAi induction, eventually becoming the dominant subpopulations, with the abnormal phenotypes "close" and "misplaced" reaching 78%. These results suggest that kinetoplast positioning is affected as a direct or indirect consequence of TbKINX1B depletion in PCF. Interestingly, TbKINX1B signal was localized to the BB of the detached new flagellum in TbBILBO1 RNAi -induced cells suggesting that BB localization of TbKINX1B does not depend on TbBILBO1 (Fig. 4C). The converse is also true as TbBILBO1 localization and expression were not affected in TbKINX1B RNAi -induced cells (Figs. 4C, S2).

TbKINX1B is essential in bloodstream cells
We further investigated the consequences of TbKINX1B RNAi down-regulation (TbKINX1B RNAi ) in BSF (Fig. 5). Quantification of BSF immunoblots (n = 3) probed with anti-TbKINX1B indicated that after 24 h of RNAi induction, TbKINX1B levels had decreased by 60% as compared to non-induced cells, and after 48 h, only 24% could be detected (Fig. 5A). Very shortly after induction, a growth defect was observed, leading to growth arrest after 24 h of induction (Fig. 5A, graph). Immuno-labelling of the flagellum, using an anti-paraflagellar rod antibody, and DAPI staining showed that RNAi-induced cells displayed an abnormal number of K and N after 24 h of induction, associated with multi-flagellated phenotypes such as 2K and 4 flagella (Fig. 5B), suggesting a defect in cell division. Quantification (n = 3) of the different cell populations showed an increase of xKxN cells over a period from 24 h to 72 h, corresponding to 14% and 38%, respectively (Fig. 5B,  bar graph). Finally, TbKINX1B RNAi -induced cells exhibited an enlarged flagellar pocket phenotype which appears to be similar to the "Big Eye" phenotype first observed after RNAi knockdown of clathrin in BSF, leading to a defect in endocytosis but not exocytosis [2]. This TbKINX1B RNAi -induced "Big Eye" also suggests several rounds of unsuccessful endocytosis, or an imbalance between endocytosis and exocytosis rates. This is also accompanied by numerous abnormal cytoplasmic vesicles, and abnormal material within the FP, as observed by electron microscopy on thin sections of whole cells (Fig. 5C-f).
The protein appears to localize to some degree along the MTQ in both PCF and BSF. This localization is differentially distributed according to the cell cycle stage with the BB + FPC signals being more abundant in 2K1N cells. This variation is probably not related to expression levels as this does not change during the cell cycle [16] and this may reflect the importance of the protein in the stages where organelle division and segregation are most important [26,38]. Kinesin motor domains are predominately associated with microtubules; their interaction with adaptor and scaffold proteins may define the type of cargo to transport and thus its function in intracellular transport. Previous FP structural analysis in T. brucei by electron tomography freeze-fracture shows that the endocytic markers (fluid or receptor-mediated) are located at the FP membrane, at the posterior and anterior face, except in the region directly associated with the MTQ, the neck MT, and the axoneme [22]. It has also been demonstrated that when internalization is blocked, in BSF, endocytic markers accumulate in a channel that runs the length of the neck and is closely associated with the MTQ, by which extracellular components can gain access to the FP lumen [22]. If we consider the motor domain homology of TbKINX1B to KIF3A (a well-studied mammalian kinesin), it suggests that the protein associates with microtubules that are linked to endosomes [7] or with recycling endosomal tubules [19]. The general endomembrane network disorganization in BSF TbKINX1B RNAi cells observed by EM suggests that TbKINX1B may play a role in intracellular trafficking, which is extremely important in BSF.

TbKINX1B is a TbBILBO1 protein partner
Immuno-electron labelling of TbKINX1B also revealed a mature and immature basal body localization. Labelling was also observed on the FPC where it colocalizes with TbBILBO1. This co-localization suggests that BILBO1 could interact with TbKINX1B as a cargo. This is supported by the immunoprecipitation assays (Fig. 3A) and the Y2H assays with a direct interaction between TbKINX1B B1BD and TbBILBO1 [21]. Further, both the TbKINX1B B1BD and the BILBO1 calcium-binding EF-hands domains are necessary for the interaction. Interestingly, TbBILBO1 EF-hands are also involved in the interaction with the newly identified FPC protein  TbBILBO2 [28]. However, unlike BILBO2 [21], the mutation of both EF-hands 1 and 2 of BILBO1 abolished the interaction with TbKINX1B. This suggests that TbKINX1B interaction with TbBILBO1 could depend on the calcium loading status of the TbBILBO1 EF-hands, unlike the interaction between the CaBP2933 EF-hands and kinesin-3 in Giardia intestinalis where binding was not calcium dependent [4].
Because TbKINX1B and TbBILBO1 physically interact, it was surprising to observe that knockdown of TbKINX1B or overexpression of its B1BD did not affect the localization of TbBILBO1, suggesting that TbBILBO1 is not an in vivo TbKINX1B cargo. This could be due to numerous reasons such as incomplete depletion of TbKINX1B by RNAi, temporally limited interaction or the requirement of numerous co-factors. Additionally, for each cargo, there may be more than one motor protein, which could indicate functional redundancy as previously described [8,14,43]. Indeed, among the 28 identified T. brucei kinesin genes [44], at least 7 (Tb927.3.2040, Tb927. 6.1770 (KIN-G), Tb927.6.2880, Tb927.7.3000, Tb927.8.4840, Tb927.11.5300 (KIN13-3), and Tb11.v5.0819) are localized at the BB-FPC area by the TrypTag Genome-wide Protein Localisation Resource [18]. One could imagine that in the absence of TbKINX1B, TbBILBO1, which is an essential protein, could be transported by a different, not yet identified, kinesin.

TbKINX1B is not essential in PCF but is vital in BSF
Long-term knockdown of TbKINX1B in PCF parasites generated BBs positioning defect, with 60% of the population showing kinetoplast positioning different to WT cells. Since the phenotype was not lethal, it suggests either that in PCF TbKINX1B has a minor function, or that RNAi was incomplete and can be compensated by other kinesins, or that the kinetoplast mis-positioning may be a downstream effect. Since PCF and BSF separate their basal bodies slightly differently, the role of TbKINX1B in basal body placement may be more important in BSF. It is also possible that TbKINX1B has a completely different role in BSF cells and that this function is disrupted once the protein is depleted. Overexpression of different TbKINX1B domains did not induce dominant-negative phenotypes. However, it did show that TbKINX1B B1BD localizes only to the FPC, suggesting that TbKINX1B binds directly to BILBO1 in vivo, supporting the Y2H and IP data. Furthermore, and as expected for a kinesin, the TbKINX1B motor domain co-localizes on MTs in vivo. It is thus possible that by simultaneously binding to BILBO1 and microtubules, TbKINX1B could transport BILBO1 along MTs.
Depletion of TbKIN-C in PCF induced a basal body segregation defect and interrupted correct cytokinesis, eventually leading to cell death [27]. Interestingly, TbKIN-C can also influence protein expression of the orphan kinesin TbKIN-D [43]. Thus, TbKINX1B could also influence the expression of other kinesins and/or cell cycle proteins that can eventually compensate for TbKINX1B (including different family groups), or vice-versa, and thus also produce secondary phenotypes related to cell division in PCF.
Clearly, TbKINX1B has a more important role in BSF. BSF depletion of TbKINX1B resulted in growth arrest after 24 h and proved to be lethal within 72 h. Cells were unable to divide correctly and displayed enlarged flagellar pockets, a phenotype known as "Big Eye". This phenotype has been previously observed in mutants with defects in endo-or exocytosis. Interestingly, it has also been observed in RNAi knockdown of some proteins belonging to the cytoskeleton, such as MORN1 and BHALIN. Here it is thought the "Big Eye" outcome is a downstream effect of interference in the structure or function of the flagellar pocket or components of that region [2,11,23,35]. It is unclear whether TbKINX1B "Big Eye" phenotypes are due to direct influence on endocytosis or due to a more general disruption of the flagellar pocket region. The difference between the phenotypes produced in PCF and BSF could be due to the efficiency of RNAi, but also to the shorter doubling time in BSF and the increased rate of endoand exocytosis in BSF compared to PCF (for a recent review, see [31]). However, the importance of the more rapid endoexocytotic system in BSF is clearly emphasized by the lethality induced by TbKINX1B RNAi.
It is important to note that the cargos for TbKINX1B, if any, remain unknown. Furthermore, TbKINX1B and BILBO1 interact in vivo but there is no evidence of BILBO1 localization being affected by depletion of the TbKINX1B in PCF and BSF. This is especially interesting in BSF given the dramatic RNAi phenotype observed. This may be explained if TbKINX1B has a structural role in BSF rather than transporting cargos; its knockdown would lead, similarly to MORN1 and BHALIN RNAi, to the Big Eye phenotype in BSF.
In conclusion, we have described the localization, the essentiality, and the plausible functional roles of the protein TbKINX1B in T. brucei. Further experiments will be necessary to understand the extent of TbKINX1B function, but our data suggest that TbKINX1B and TbBILBO1 are, during a restricted temporal period, at least part of a protein complex. the Aquitaine Regional Council Grant -20111301014, the ANR (ANR-09-BLAN-0074 and ANR PRCI ANR-20-CE91-0003), and the Laboratoire d'Excellence (LabEx) ParaFrap grant (ANR-11-LABX-0024). DP was a postdoc recipient from the LabEx ParaFrap. The electron microscopy was done in the Bordeaux Imaging Center, a service unit of the CNRS-INSERM and Bordeaux University, member of the national infrastructure France BioImaging supported by the French National Research Agency (ANR-10-INBS-04).

Supporting information
The Supplementary materials of this article are available at https://www.parasite-journal.org/10.1051/parasite/2022015/olm Figure S1: TbKINX1B can hydrolyse ATP in vitro. A. ATPase activity using the ELIPA in vitro assay.