A novel function for the atypical small G protein Rab-like 5 in the assembly of the trypanosome flagellum

The basic structure of cilia and flagella is the axoneme, a set of nine doublet microtubules organised in a cylindrical manner with (9+2) or without (9+0) a central pair of microtubules. Dynein arms and radial spokes are involved in generation and control of beating and are mostly found on 9+2 axonemes (Ralston and Hill, 2008). Ciliary microtubules originate and extend from the basal body, a cytoplasmic centriolar-like structure that is usually composed of nine triplet microtubules. During cilium construction, proteins are added at the distal tip of the elongating organelle by a process called intraflagellar transport (IFT) (Rosenbaum and Witman, 2002). IFT is defined as the bidirectional movement of protein complexes from the base of the cilia to the distal tip (anterograde transport, powered by heterotrimeric kinesin II) and back towards the cell body (retrograde transport, driven by a specific dynein complex). IFT complex A (at least six polypeptides) and complex B (at least 11 polypeptides) (Cole et al., 1998) are required for retrograde and anterograde IFT, respectively. IFT proteins and their functions are conserved in all flagellated species examined so far: algae, nematodes, insects, trypanosomes, ciliates and mammals (Kohl and Bastin, 2005). These IFT proteins are rich in protein-protein interaction domains such as WD40 and tetratricopeptide repeats (TPR) or coiled-coils (Cole, 2003). One of few exceptions is the small G protein IFT27 (RAB-like 4), which is part of the B complex and is required for flagellum formation in Chlamydomonas (Cole et al., 1998; Qin et al., 2007). Other small G proteins have been shown to participate to formation of cilia and flagella such as ARL3 in Leishmania (Cuvillier et al., 2000), ARL6 [a protein that is mutated in some individuals who suffer from the Bardet-Biedl syndrome, a cilia-related disease (Fan et al., 2004)] and Rab8 in mammalian cells (Omori et al., 2008).

Rab-like 5 (RABL5 or IFTA-2 in C. elegans) is an atypical G protein that lacks the G4 domain of the GTP-binding region and the prenylation motif. The protein has been localised to the basal body and the ciliary compartment in C. elegans sensory neurons and to the primary cilium of inner medullary collecting duct (IMCD) cells or of retinal pigmented epithelial (RPE1) cells (Schafer et al., 2006; Yoshimura et al., 2007). RABL5 fused to GFP is transported in the nematode cilia but RABL5 inactivation in two independent deletion mutants did not result in visible alteration of ciliary structure, nor in perturbation of sensory perception in the mutant worms. However, an extended lifespan phenotype was observed and was linked to defects in DAF2/insulin/IGF-like 1 signalling pathway, leading to the suggestion that RABL5 is involved in sensing rather than in axoneme construction (Schafer et al., 2006).

Using comparative genomics of species assembling flagella with or without IFT, we and others have identified several genes that are restricted to organisms assembling their flagella by IFT (putatively involved in intraflagellar transport or PIFT) (Absalon et al., 2008b; Avidor-Reiss et al., 2004; Briggs et al., 2004; Li et al., 2004). The RABL5 gene was found to belong to this category, suggesting a function related to IFT. However, it could not be identified in insect genomes such as Drosophila or Anopheles, despite the fact that cilia in their sensory neurons (9+0) are assembled by IFT. If RABL5 is involved in sensing functions (Schafer et al., 2006), its exact role remains to be determined as it is present in organisms with very different types of cilia and the range of signals that these cilia might have to detect are likely to diverge extensively. We therefore investigated the role of RABL5 in a different organism: the protist Trypanosoma brucei. This micro-organism possesses a single flagellum with a classic motile 9+2 axoneme and has become an attractive model for functional studies (Baron et al., 2007; Broadhead et al., 2006; Kohl and Bastin, 2005; Ralston and Hill, 2008). The trypanosome RABL5 was found to localise to the flagellum and basal body compartments, where it demonstrated the characteristics of an IFT protein similar to that observed in C. elegans. However, knocking-down RABL5 by RNAi resulted in the formation of very short flagella that were filled with IFT-like material. These data, thus, reveal a novel function for RABL5 in retrograde IFT. In contrast to the well-conserved function of IFT proteins in flagellum construction, RABL5 is the first matrix ciliary protein to show radically different roles between two ciliated/flagellated organisms.

The RABL5 gene (GeneDB Accession Number, Tb11.01.8590) is 660 nucleotides long and encodes a predicted protein of 219 amino acids long with a calculated molecular mass of 24.4 kDa and an iso-electric point of 5.2. Classification of RABL5 as a putative Rab protein is based on BLAST searches and on sequence alignment of the G-domains found in most Rab proteins and involved in GTP/GDP binding. The first three G-domains of the five Ras GTPase consensus sequences are well-conserved in all the RABL5 sequences examined (Fig. 1). By contrast, a conventional G4 box is not present in all the RABL5 analysed, whereas the G5 box shows more diversity. None of the RABL5 proteins exhibits the consensus sequence E(T/G/S)(G/S)A, the closest match being found in Trypanosomatids (T. brucei, T. cruzi, Leishmania major, L. braziliensis and L. infantum). In these five species, the consensus sequence is ETSL. A variation in the consensus sequence is found in two other protists: DTSI in both Paramecium orthologues and KTSM in Giardia lamblia. All vertebrate RABL5 sequences display the sequence HSNL (human, mouse, rat, dog, cow, opossum and fish), whereas the most divergent sequences are found in Chlamydomonas (AINP) or Caenorhabditis (AAHF) species and do not fit with the consensus. The C. elegans sequence contains two insertions of 23 and 25 amino acids before and after the G1 domain, respectively, making it the longest of the RABL5 sequences identified (252 amino acids compared with 185-219 amino acids for the others). The Trypanosomatid RABL5 proteins are slightly larger than their vertebrate counterpart owing to the insertion of six amino acids between the G1 and G2 domains and 10 amino acids between the G3 and the G5 domains. Additionally, it possesses a longer C-terminal extension. The prenylation domain CCxx xCCx, which is usually located at the C-terminal end of the Rab proteins, is missing, indicating that they are unlikely to be lipid modified. A clear homologue of RABL5 was found in all ciliated/flagellated organisms except in Drosophila and Anopheles. Like all the well-characterised IFT genes, RABL5 is missing from the genome of Plasmodium falciparum, a parasite that assembles its flagella in the cytoplasm (Briggs et al., 2004). Finally, BLAST analysis demonstrated that RABL5 proteins are clearly distinct from all other RAB families, including RABL4/IFT27 (T. brucei RABL4 and RABL5 are less than 15% identical).

To determine the location of the trypanosome RABL5, the full-length RABL5 was fused to glutathione-S-transferase (GST), expressed in bacteria, purified and injected into mice to generate polyclonal antibodies. Control mice were immunized with GST alone and sera were adsorbed against GST prior to being used for western blotting. In wild-type cells, the antibody detected a single band migrating close to the 25 kDa marker, i.e. in the range of the expected molecular weight of RABL5 (Fig. 2A, arrow). In cells expressing a GFP::RABL5 fusion protein, a supplementary band of M_r ∼50 K was visible, which is comparable with the predicted mass of the fusion protein (Fig. 2A, star; and see below). The anti-RABL5 antiserum was used in indirect immunofluorescence assay (IFA) in combination with the monoclonal antibody MAb25 that stains exclusively a cytoskeletal component of the axoneme (Absalon et al., 2007; Pradel et al., 2006) to reveal the flagellum (Fig. 2B, arrows), and with DAPI to stain DNA in the nucleus and the mitochondrial genome (kinetoplast). The kinetoplast is physically linked to the basal body of the flagellum and hence is a good indicator of basal body position (Robinson and Gull, 1991). Trypanosomes maintain their existing flagellum while growing a new one; hence, cells can be found with one (Fig. 2B, lower cell) or two flagella (upper cells at Fig. 2B). The new flagellum (Fig. 2B, yellow arrows) can easily be identified as it is always positioned at the wider, posterior, end of the cell (Sherwin and Gull, 1989). The anti-RABL5 antiserum decorated equally both old and new flagella but the labelling was discontinuous and slightly shifted from the central axis of the flagellum when compared with that observed for MAb25 (Fig. 2B, merged panel). The anti-RABL5 signal extended to the basal body region where it often appeared brighter (Fig. 2B, arrowheads). By contrast, only a weak cytoplasmic background labelling was produced using the control anti-GST (data not shown). We also generated trypanosomes expressing GFP-tagged RABL5 in a wild-type background. On both live and fixed cells, GFP::RABL5 localised to the cytoplasm, basal body and flagellum (Fig. 2C; data not shown), i.e. a similar pattern to the results from IFA experiments where cells were probed with the anti-RABL5 and further supports an association of RABL5 with the flagellum and basal body compartments.

To investigate the localisation pattern linked to the basal body, we performed double IFA labelling with MAb22, a monoclonal antibody marker of the fibres that link the proximal part of both the mature and pro-basal body to the kinetoplast (Bonhivers et al., 2008). In all cells, the MAb22 and the RABL5 (with the antibody or the GFP fusion) signals exhibited close but clearly different locations, with RABL5 being always on the most distal region of the basal body (Fig. 2C), a localisation similar to that of the three known IFT proteins in trypanosomes (Absalon et al., 2008b). To further establish the association of RABL5 with the IFT particles, detergent treatment was applied to trypanosomes, a procedure that removes the cell membrane and extracts an intact cytoskeleton (Sherwin and Gull, 1989), including the flagellum (Broadhead et al., 2006). Under these conditions, RABL5 fractionated with the soluble fraction, similar to the endoplasmic reticulum control marker BiP and in contrast to cytoskeletal proteins such as the PFR (Fig. 2D). Analysis of cytoskeletons by IFA showed that RABL5 (data not shown) or the GFP::RABL5 fusion protein (Fig. 2E) only remained associated to the basal body upon detergent treatment, whereas the flagellum and cell body signal was abolished, as reported previously for IFT proteins (Absalon et al., 2008b). The absence of RABL5 in the cytoskeletal fraction probably reflects the low amount of proteins present at the basal body compared with the flagellum and the cell body.

To firmly establish the association between IFT proteins and RABL5, double IFA staining was performed with antibodies against IFT172 (involved in anterograde transport, Fig. 3A) or against PIFTF6/IFT144 (involved in retrograde transport, Fig. 3B) (Absalon et al., 2008b). A close but not exact colocalisation of the RABL5 signal at the basal body and within the flagellum matrix was observed in both experiments (Fig. 3A,B). Taken together, these results suggest that RABL5 could be associated to IFT.

Trypanosomes expressing GFP::RABL5 were used to visualise the possible movement of RABL5 in the flagellum of live cells. Clear intraflagellar movement of GFP::RABL5 was detected, mainly from the base of the flagellum to the tip (anterograde movement) (Fig. 4A; supplementary material Movie 1). As observed for GFP::IFT52 in trypanosomes (Absalon et al., 2008b) or with GFP::IFT20 in mammalian primary cilia (Follit et al., 2006), retrograde events were more difficult to visualise. We next investigated the behaviour of GFP::RABL5 in cell lines where IFT had been perturbed. First, when retrograde IFT is blocked, trypanosomes assemble short flagella filled with IFT particles that cannot be recycled to the basal body region (Absalon et al., 2008b). Whereas RABL5 exhibited the usual basal body and flagellum signals in non-induced cells (Fig. 4B), it accumulated in the short flagella of the IFT140^RNAi mutant in every cell examined (Fig. 4C,D). All the IFT proteins examined to date accumulate in the short flagella of retrograde transport mutants, in contrast to the proteins from the axoneme or the PFR (Absalon et al., 2008b). Inhibition of anterograde movement in the IFT88^RNAi mutant prevents flagellum formation. Under these conditions, RABL5 was found only at the basal body and accumulation of large amounts of material was not observed (Fig. 4E). These data demonstrate that RABL5 exhibits intraflagellar movement in live cells and behaves like the IFT proteins both in normal cells and in retrograde and anterograde IFT mutants.

Tetracycline-inducible RNA interference (RNAi) was used to deplete RABL5 and examine its functions in trypanosomes. RT-PCR confirmed that RABL5 mRNA was rapidly degraded upon induction of RNAi silencing, whereas the level of mRNA for aldolase used as control was not affected (Fig. 5A). Western blot analysis using the anti-RABL5 polyclonal antibody showed that the endogenous RABL5 dropped rapidly and became barely detectible 2 days after induction (Fig. 5B). RABL5^RNAi cells grew normally for 4 days until they stopped proliferating and died (Fig. 5C). Examination of cells 3-4 days after induction revealed the emergence of short cells that were typical of inhibition of flagellum formation (Absalon et al., 2008b; Absalon et al., 2007; Kohl et al., 2003) (Fig. 5D). IFA with the axoneme marker MAb25 revealed that RABL5^RNAi cells induced for 4 days assemble abnormally short flagella (Fig. 5D), the average length of which was 3.1±1.4 μm (n=50) instead of 18.5±2.9 μm (n=50) for non-induced controls. Trypanosomes assemble a new flagellum while maintaining the existing one (Fig. 5E) and scanning electron microscopy revealed that RABL5^RNAi cells induced for 3 or 4 days still possessed an apparently intact old flagellum (Fig. 5F,G, white arrows) but assembled a very short new flagellum (yellow arrows). After cytokinesis, the cell inheriting the short new flagellum was too small and its flagellum did not show signs of further elongation (Fig. 5H,I, arrows). Supplementary membrane-like extensions were frequently observed (Fig. 5F-J, arrowheads), presumably corresponding to the `flagellar sleeve' reported in other IFT mutants in trypanosomes (Absalon et al., 2008b; Davidge et al., 2006). Some cells were also observed without a flagellum but structures resembling vesicles could be recognised emerging from the flagellar pocket area (Fig. 5J).

This phenotype strongly resembles that observed upon inhibition of retrograde IFT. This inhibition leads to formation of short flagella filled with IFT proteins but these proteins cannot be recycled to the cell body (Blacque et al., 2006; Pazour et al., 1999; Porter et al., 1999; Signor et al., 1999). We therefore analysed the location of IFT proteins in RABL5^RNAi cells. IFA with the anti-IFT172 antiserum revealed normal staining in non-induced cells (Fig. 6A), whereas clear accumulation of IFT172 in the short flagella of RABL5^RNAi cells induced for 4 days was striking (Fig. 6B). Quantitative analysis confirmed the presence of a large proportion of these cells similar to retrograde IFT140^RNAi or DHC1b^RNAi cells, but in contrast to the anterograde IFT88^RNAi mutant (Fig. 6C).

To analyse the nature of the material accumulated in the short flagella of RABL5^RNAi trypanosomes, we examined these cells by transmission electron microscopy. Whereas control samples presented the normal aspect of the basal body and the flagellum emerging from the flagellar pocket (Fig. 7A), RABL5^RNAi cells exhibited bulbous short flagella filled with IFT-like material, a feature typical of cells with defective retrograde transport (Fig. 7B,C). Quantitative analysis revealed that 59% of the sections through flagella of the RABL5^RNAi mutant cells showed clear accumulation of IFT material (n=27), a proportion similar to that observed for the retrograde transport mutants DHC1b^RNAi (35%, n=48) and IFT140^RNAi (45%, n=38). By contrast, no accumulation of IFT-like material could be detected in the short flagellar stubs of the anterograde transport mutants IFT88^RNAi (0%, n=23) and IFT172^RNAi (0%, n=28). Furthermore, RABL5^RNAi cells also exhibited all the typical phenotypes that resulted from defects in flagellum formation: inhibition of cytokinesis, growth arrest and multiple rounds of basal body duplication that fail to migrate and also defects in formation of the flagellar pocket (supplementary material Fig. S1; data not shown) (Absalon et al., 2008a; Absalon et al., 2008b; Absalon et al., 2007; Davidge et al., 2006; Kohl et al., 2003).

The endogenous RABL5 or the fusion protein GFP::RABL5 show a clear localisation to the basal body and the flagellum matrix. This pattern is very similar to what was observed for IFT proteins such as IFT20, IFT52 and IFT172 (Absalon et al., 2008b). The GFP::RABL5 fusion protein is transported in the flagellum matrix and behaves similarly to the GFP::IFT52 protein (Absalon et al., 2008b). These results are in agreement with the presence of RABL5 in sensory cilia of C. elegans where it also undergoes IFT (Ou et al., 2007; Schafer et al., 2006). At least two functions can be considered for RABL5: (1) it is a cargo transported by the IFT particles; or (2) it is an integral component of the IFT particle. In Chlamydomonas, RABL5 (FAP9) has been found in the flagellum matrix (Pazour et al., 2005) and its mRNA is upregulated upon deflagellation (Li et al., 2004), similar to the expression profile of other flagellar proteins. By contrast, RABL5 is not associated to the cytoskeletal fraction of the flagellum, neither in the green algae (Pazour et al., 2005), nor in trypanosomes (Broadhead et al., 2006). Finally, RABL5 was not found among the proteins present in IFT particles purified from Chlamydomonas (Cole, 2003).

In mammalian cells, RABL5 has been localised to the basal body and the ciliary compartment of the primary cilium in inner medullary collecting duct (IMCD) cells or in retinal pigmented epithelial (RPE1) cells (Schafer et al., 2006; Yoshimura et al., 2007). It has also been found at the Golgi level in non-ciliated NIH3T3 cells (Yang et al., 2007). Intriguingly, we noticed that in trypanosome cells overexpressing large amounts of GFP::RABL5, the protein was consistently found at the Golgi apparatus in addition to the flagellum and basal body (data not shown). It is difficult to establish whether this localisation is the result of overexpression of the fusion protein or whether the absence of RABL5 at the Golgi level in wild-type cells is due to a lack in sensitivity of the anti-RABL5 antibody. In several mammalian cell lines, IFT20 is present at the Golgi apparatus in addition to the basal body and the matrix of the flagellum, suggesting it might function in the delivery of flagellar membrane proteins (Follit et al., 2006). However, this was not observed in trypanosomes where IFT20 is exclusively localised to the flagellum and the basal body (Absalon et al., 2008b).

RNAi knock-down demonstrates that RABL5 is required for flagellum formation and produces a phenotype typical of defects in retrograde IFT, similar to those observed upon silencing of IFT122, IFT140 (Absalon et al., 2008b) or the retrograde transport motor DHC1b (Kohl et al., 2003). This phenotype supports the hypothesis that RABL5 is part of the IFT particle or somehow regulates its function. It is possible that RABL5 association with the IFT particle(s) is less stable or only transient, or its stoichiometry is lower compared with the other IFT component proteins identified thus far. Interestingly, a better colocalisation of RABL5 has been obtained with PIFTF6/IFT144 (required for retrograde IFT) compared with IFT172 (required for anterograde IFT). However, this could also be due to the access of antibodies to their target, especially if these proteins are part of large complexes as observed in the green algae Chlamydomonas. In any case, this situation is in stark contrast to C. elegans, where RABL5 demonstrates IFT but is not required for formation of cilia (Schafer et al., 2006). RABL5 is the first reported matrix ciliary protein that performs such drastically different functions between two organisms. In molecular terms, this difference in function could be related to sequence divergences. For example, the C. elegans RABL5 protein possesses two long insertions that are not found in other species and does not contain a G5 domain, in contrast to Trypanosomatids where three out of four residues of the G5 box are conserved. At the biological level, one possible explanation could come from the different types of cilia between trypanosomes and nematodes. C. elegans possesses 9+0 non-motile cilia, whereas T. brucei possesses a 9+2 motile flagellum that contains a central pair, dynein arms and radial spokes. These multi-protein complexes are transported by IFT (Piperno et al., 1996; Qin et al., 2004) and this requirement could modify the organisation of the periphery of the IFT particles. If RABL5 is positioned differently in the IFT particle in various types of cilia, its absence could have different consequences on IFT and flagellum formation. Understanding the exact role of RABL5 will require knowledge of the biochemistry of the IFT particles that is not yet available neither for T. brucei nor for C. elegans.

Despite their well-conserved structure, cilia and flagella have become specialised to fulfill variable functions. The best example is found in mammals, where multiple types of cilia are present: motile 9+2 cilia in the respiratory epithelium, motile 9+0 cilia at the embryonic node, non-motile 9+0 primary cilia, the connecting cilium of photoreceptors or the motile sperm flagellum all have different organisations and roles. It was long-assumed that the function of IFT proteins was fully conserved in different cilia, but recent data show subtle differences for motor or IFT proteins in C. elegans or in zebrafish (Krock and Perkins, 2008; Mukhopadhyay et al., 2007; Scholey, 2008). This diversity could be especially important in complex genetic diseases related to defects in cilia and flagella such as in the Bardet-Biedl syndrome (Marshall, 2008). Extensive symptom diversity between individuals has been reported and mouse models have revealed that mutations in genes involved in Bardet-Biedl syndrome did not have the same structural consequences on all types of cilia (Davis et al., 2007). The results reported here show that radical phenotypic variation can be observed for a conserved ciliary protein and emphasise the importance of analysing carefully the function of these proteins in different contexts and in different model organisms.

The T. brucei RABL5 protein sequence (XP_829740.1 or Gene DB id Tb11.01.8590) was compared with that of several other organisms: T. cruzi, the South American trypanosome, responsible for Chagas disease XP_805503.1; Leishmania major (XP_001685647.1), L. infantum (XP_001468015.1) and L. braziliensis (XP_001567701.1); the divergent protist Giardia lamblia (XP_001708211.1); the ciliate Paramecium tetraurelia, in which two different protein-coding genes were identified (XP_001458678.1 and XP_001461610.1); the green algae Chlamydomonas reinharditii (XP_001689669.1); the nematodes C. elegans (NP_503073.1) and C. briggsae (XP_001666841.1); and several vertebrates such as human (NP_073614.1), mouse (NP_080349.1 and AAH09150.1), rat (NP_001011902.1), dog (XP_536853.2), cow (NP_001015612.1), opossum (XP_001366207.1) and zebrafish (NP_001017884.1).

T. brucei cell lines were derivatives of strain 427 and grown in SDM79 medium with hemin and 10% foetal calf serum. The 29-13 cell line expressing the T7 RNA polymerase and the tetracycline-repressor (Wirtz et al., 1999) has been described previously, as well as the RNAi cell lines IFT88^RNAi (Kohl et al., 2003), IFT140^RNAi and IFT172^RNAi (Absalon et al., 2008b).

For generation of the RABL5^RNAi cell line, a 447 nucleotide fragment of RABL5 was cloned in the pZJM vector (Wang et al., 2000), allowing for tetracycline-inducible expression of dsRNA generating RNAi upon transfection in the 29-13 recipient cell line. The dsRNA is expressed from two tetracycline-inducible T7 promoters facing each other in the pZJM vector. Primers were selected using the RNAit algorithm to ensure that the fragment lacked significant identical residues to other genes to avoid cross-RNAi (Redmond et al., 2003). The fragment was amplified by PCR using primers CGATCGAAGCTTGGATAGTCTTTTGCTTC and GTCATCTCGAGCGCGGATAG TCTTTTGCTTC from T. brucei genomic DNA, digested with HindIII and XhoI (restriction sites underlined) and ligated in the corresponding sites of the pZJM vector. The pZJMRABL5 plasmid was linearized with NotI and transformed in 29-13 cells. Transfected cells were immediately cloned and antibiotic-resistant cell lines obtained. RNAi was induced by addition of 1 μg tetracycline per ml of medium and fresh tetracycline was added at each cell dilution. For generation of GFP::RABL5, the full-length RABL5 gene was amplified by PCR with the proof-reading enzyme PfuI using GCATCAGATATCATGTCGGACGACTTGGTAAAG (EcoRV site underlined) and GCATCGGATATCTCAACGCTTCAGCCGCGGATAGTC (EcoRV site underlined) and ligated into the pPCPFR plasmid digested with EcoRV (Sabrina Absalon and P.B., Institut Pasteur, Paris, unpublished results). The eGFP gene was amplified by PCR with the proofreading enzyme PfuI using GCATCAAAGCTTATGGTGAGCAAGGGCGAG (HindIII site underlined) and GCATCGAAGCTTCTTGTACAGCTCCGTCCAT (HindIII site underlined) and then ligated into the pPCPFRRABL5 plasmid to generate pPCPFReGFPRABL5. This plasmid was linearized by NsiI and integrated in the intergenic region of PFR2. Inserted sequences and flanking regions were sequenced to confirm correct fusion (Genome express). Transfected cells were immediately cloned and all antibiotic-resistant cell lines characterized by direct fluorescence.

Full-length RABL5 amplified with the PfuI proofreading enzyme, using GCATCAGAATTCATGTCGGACGACTTGGTAAAG (EcoRI site underlined) and GCATCGAAGCTTTCAACGCTTCAGCCGCGGATAGTC (HindIII site underlined) was digested with EcoRI and HindIII, and ligated in compatible sites of the pGEXB vector. The plasmid was sequenced to confirm correct fusion with GST and transformed in E. coli BL21. Protein expression was analyzed by SDS-polyacrylamide gel electrophoresis (PAGE) followed by Coomassie staining. Glutathione transferase (GST)-coupled proteins were purified as described previously (Smith and Johnson, 1988) and 20 μg were administrated by four successive subcutaneous injections (every 3 weeks) to BALB/c mice for immunization. After bleeding, sera were absorbed against GST. Sera from mice immunized with GST alone were used as negative controls.

For IFA with anti-RABL5 antisera, intact cells or detergent-extracted cytoskeletons were washed, settled on poly-L-lysine-coated slides and fixed in methanol (–20°C) for 5 minutes. For detergent treatment, cells were settled on poly-L-lysine coated slides and exposed to 1% Nonidet P-40 in spindle stabilisation buffer [4 M glycerol, 5 mM EGTA, 10 mM MgCl₂, 10 mM PIPES and 0.1% (v/v) Triton X-100 (pH 6.5)]. Blocking was performed for 45-60 minutes in PBS containing 1% bovine serum albumin and washed slides were incubated with anti-RABL5 antisera for 45-60 minutes. Slides were washed and incubated with anti-mouse secondary antibodies coupled to FITC (Sigma). Other antibodies used are L8C4, a marker of flagellum assembly (Kohl et al., 1999); MAb22 a marker of both the mature and the pro-basal body; and MAb25, which recognises a protein found all along the axoneme (Bonhivers et al., 2008; Absalon et al., 2007; Pradel et al., 2006) and mouse polyclonal antisera raised against a GST::IFT172 or a GST::PIFTF6 fusion protein (C.A., T.B., G.T., E.D. and P.B., unpublished). GFP was observed directly or upon IFA using an anti-GFP antibody (Invitrogen). Subclass-specific secondary antibodies coupled to FITC (Sigma), Alexa 488 or Alexa 594 (Invitrogen), Cy3 or Cy5 (Jackson) were used for double labelling. For visualization of GFP, live cells were mixed with a solution of 3% LMP agarose (Biorad) to reduce cell movement. Samples were observed with a DMR Leica microscope and images were captured with a Cool Snap HQ camera (Roper Scientific). Images were analysed using the IPLab Spectrum 3.9 software (Scanalytics & BD Biosciences).

Transmission (Branche et al., 2006) and scanning (Absalon et al., 2007) electron microscopy were carried out exactly as previously described.

Total RNA was extracted using Trizol from cells grown with or without tetracycline for the indicated periods of time. DNA was eliminated by DNase treatment and RNA purity was confirmed by conventional PCR. For determination of optimal conditions for RT-PCR, 10-2000 ng template total RNA from wild-type trypanosomes were incubated with 0.8 μM primers and amplified by the mean of the RT-Taq Platinum (Invitrogen) enzyme mix to find out the linear range of the reaction. Appropriate temperature for annealing was determined using a temperature gradient on a DNA Engine (BioRad) PCR machine. Once these parameters were established, 100 ng of RNA from induced and non-induced RABL5^RNAi cells was used for semi-quantitative RT-PCR performed as described (Durand-Dubief et al., 2003). To assess the extent of RNAi silencing in induced samples, up to 2 μg RNA were used. Primers used were ATGTCGGACGACTTGGTAAAG (forward) and TCAACGCTTCAGCCGCGGATAG (reverse). They are outside the region selected for dsRNA and encompass the full coding sequence.

Cells (1×10⁶/μl) were washed in PBS and boiled in gel sample buffer before SDS-PAGE. Proteins were transferred to PVDF membranes and incubated with a 1:500 dilution of anti-RABL5. Membranes were stripped and probed with the anti-PFR antibody L13D6 (Kohl et al., 1999) or with a rabbit anti-Bip antibody (Bangs et al., 1993) as loading control and revealed with ECL+ (Amersham).