IFT25 is required for the construction of the trypanosome flagellum

Cilia and flagella are conserved across eukaryotes and perform variable functions. Intraflagellar transport (IFT) is the movement of protein particles or trains that deliver precursors at the tip of the organelle for assembly (Craft et al., 2015; Kozminski et al., 1993). Trains are polymers made of two protein complexes termed IFT-A and IFT-B (Cole et al., 1998; Piperno and Mead, 1997; Taschner and Lorentzen, 2016). Inhibition of IFT blocks cilia construction in all species investigated to date. In general, absence of an IFT-A protein interferes with retrograde transport, leading to the formation of short cilia that accumulate IFT proteins whereas absence of an IFT-B protein results in failure to construct the organelle (Absalon et al., 2008b; Blacque et al., 2006; Efimenko et al., 2006; Iomini et al., 2009).

An exception to the IFT conservation is IFT25 (also known as HSPB11 in mammals), which is missing from the genome of 15 ciliated species (van Dam et al., 2013). IFT25 was identified as a candidate chaperone protein based on some sequence homology with heat-shock proteins (Bellyei et al., 2007). However, this was not supported by structural analysis of the Chlamydomonas IFT25 protein (Bhogaraju et al., 2011). IFT25 associates with the small G protein IFT27 and is added to the IFT-B complex prior to IFT activation (Bhogaraju et al., 2011; Taschner et al., 2014; Wang et al., 2009). The absence of IFT25 in multiple species raises the question of its actual role. In contrast to other IFT genes, mice with deletion of the IFT25-encoding gene are viable until birth. Cilia in the lung and in the trachea are present and exhibit a normal morphology, suggesting that the protein is not essential for cilia assembly. Embryonic fibroblasts of the Ift25 mutant contained all examined IFT proteins with the exception of IFT27 (Keady et al., 2012). Nevertheless, embryos died at birth with multiple defects related to dysfunctions in hedgehog signalling. The distribution of components of this signalling pathway was drastically modified, leading to a model where IFT25 and IFT27 would contribute to the transport of signalling molecules via the BBSome complex (Eguether et al., 2014; Keady et al., 2012; Liew et al., 2014). Accordingly, a point mutation in the human IFT27 gene has been shown to result in Bardet–Biedl Syndrome (Aldahmesh et al., 2014). By contrast, a predicted biallelic loss of IFT27 led to a much more severe ciliopathy in a foetus (Quelin et al., 2018). Intriguingly, IFT25 is conserved in algae and multiple protists where the hedgehog cascade is absent, raising the question of its role in these organisms. Inhibition of IFT25 expression in Chlamydomonas did not visibly affect either IFT or flagellum construction but perturbed BBSome trafficking (Dong et al., 2017).

In an effort to clarify the function of IFT25, we investigated its function in the protist Trypanosoma brucei, another model for flagellum construction (Vincensini et al., 2011). The 22 IFT genes are conserved in T. brucei, IFT trafficking was demonstrated and inhibition of IFT gene expression severely impedes flagellum construction (Absalon et al., 2008b; Blisnick et al., 2014; Buisson et al., 2013; Davidge et al., 2006; Franklin and Ullu, 2010; Kohl et al., 2003). Here, we show that the trypanosome IFT25 is a bona fide member of the IFT-B complex, that it undergoes IFT trafficking in interaction with IFT27, and that it is essential for retrograde transport.

In T. brucei, IFT25 was detected in purified intact flagella of both insect (Subota et al., 2014) and bloodstream (Oberholzer et al., 2011) stages but not in detergent and salt-extracted flagellar fractions (Broadhead et al., 2006), indicative of an association to the flagellum matrix. To investigate its exact localisation, a GFP::IFT25 fusion was expressed. Western blot analysis using an anti-GFP antibody showed a single band migrating between the 37 and the 50 kDa markers, in agreement with the expected mass of the fusion protein (45 kDa) (Fig. 1A). The anti-GFP antibody was used in double immunofluorescence assays (IFA) with the anti-IFT27 antibody revealing spots along the flagellum that were stained by both antibodies (Fig. 1B).

IFT25 and IFT27 were not found in IFT-B complex purification, possibly because of their small size (Franklin and Ullu, 2010). To establish whether IFT25 was a member of the IFT-B complex, immunoprecipitation experiments were carried out on whole-cell extracts from trypanosomes expressing GFP::IFT25. Protein lysates were immunoprecipitated with either an anti-GFP antibody or an anti-aldolase as a control. Precipitates were analysed by western blotting with antibodies against either IFT22 (Adhiambo et al., 2009) or IFT27 (Huet et al., 2014), two recognised members of the IFT-B complex. The anti-GFP antibody successfully immunoprecipitated IFT22 and IFT27, in contrast to the anti-aldolase antibody (Fig. 1C). These results reveal that IFT25 is a member of the IFT-B complex in trypanosomes.

To further link IFT25 to intraflagellar transport, live cells expressing GFP::IFT25 were observed by video microscopy. Transport in the anterograde direction was easily detected, whereas retrograde transport was more discrete but clearly present upon separation (Movie 1; still images are shown in Fig. 1D). Kymograph analyses (Fig. 1E) determined a mean (±s.d.) anterograde velocity of 2.7±0.66 µm/s (n=229), while mean retrograde velocity was 3.1±0.90 µm/s (n=296). Similar values have been observed for fusion proteins with IFT52 (Buisson et al., 2013), IFT81 (Bertiaux et al., 2018; Bhogaraju et al., 2013) or IFT27 (Huet et al., 2014).

Since immunoprecipitations were performed on whole-cell extracts, we used bimolecular fluorescence complementation assays (Kerppola, 2008) to investigate where IFT25 and IFT27 interact. This approach had been used successfully for interactions between IFT proteins or IFT proteins and transition fibre components in C. elegans (Wei et al., 2013; Yi et al., 2017). We fused the N-terminal portion of YFP spanning the first 155 amino acids (aa) to IFT25 and its C-terminal portion (aa 156–239) to IFT27. Constructs were designed for endogenous tagging at the N-terminal end of each protein, ensuring appropriate control of expression (Kelly et al., 2007). Expression of each protein individually did not lead to visible fluorescence in live cells but could be detected by IFA using a rabbit polyclonal antibody against GFP (Fig. 2A,B). By contrast, fluorescent signals and typical IFT movements were easily detected in live cells when both proteins were co-expressed (Movie 2; still images are shown in Fig. 2C). We conclude that IFT25 is a genuine component of the IFT-B complex in T. brucei because: (1) it is found within the matrix of the flagellum and concentrates at its base, (2) it traffics in both anterograde and retrograde directions at a speed and frequency similar to those reported for other IFT proteins, and (3) it is associated with other IFT-B proteins as shown by immunoprecipitation and bimolecular fluorescence complementation assays.

To evaluate the role of IFT25 in T. brucei, an inducible RNA interference (RNAi) approach was used. The first phenotype of the IFT25^RNAi cell line was an obvious growth defect upon depletion of the protein (Fig. S1). Scanning electron microscopy analysis of non-induced cells showed the conventional morphology with elongated cells and a long flagellum (Fig. 3A) whereas growth of IFT25^RNAi cells in RNAi conditions for 24 (Fig. 3B) or 48 h (Fig. 3C) resulted in small cells with short flagella. IFT25^RNAi cells were fixed and prepared for transmission electron microscopy (TEM) analysis. Sections through flagella or their base exhibited their normal aspect in control non-induced IFT25^RNAi cells with a normal basal body, the transition zone and axoneme (Fig. 3D). By contrast, the flagella of induced cells were very short, and their microtubule doublets looked disorganised with a large amount of amorphous material accumulated towards the distal end (Fig. 3E–G). Small vesicles were frequently detected among this material. This is reminiscent of a defect in retrograde transport in T. brucei (Absalon et al., 2008a; Blisnick et al., 2014). A similar phenotype was reported for IFT22 (Adhiambo et al., 2009) and IFT27 (Huet et al., 2014), the two small G protein members of the IFT-B complex.

We analysed the fate of IFT proteins in the IFT25^RNAi cell line. Upon induction, no changes of the total steady-state amounts of IFT-B members IFT22, IFT27 or IFT172 were observed by western blotting (Fig. 4A). IFA using the anti-IFT27 antibody showed that the protein is found in the cytoplasm upon depletion of IFT25 and therefore fails to access the flagellum (Fig. 4B). To evaluate the inverse relationship, GFP::IFT25 was expressed in the IFT27^RNAi cell line and its distribution was investigated by IFA with the anti-GFP antibody. GFP::IFT25 was found in the flagellum in control conditions but was dispersed in the cytoplasm upon silencing of IFT27 (Fig. 4C). IFT27 is a small G protein and needs to bind GTP to be associated to the IFT-B complex and to access the flagellum (Huet et al., 2014). GFP::IFT25 was introduced in the IFT27^RNAi cell line expressing an RNAi-resistant version of IFT27 with point mutations locking this small G protein in the GDP-bound or GTP-bound form. In the case of the GTP-locked form, GFP::IFT25 can still associate to IFT trains (Fig. S2A), in agreement with the fact that this version can almost completely rescue the RNAi phenotype. By contrast, the GDP-locked version cannot complement the phenotype, does not associate to the IFT-B complex and is found in the cytoplasm and not in the flagellum. In these conditions, GFP::IFT25 is only present in the cytoplasm (Fig. S2B). These results show that the simultaneous presence of IFT25 and GTP-bound IFT27 is needed for their association to the IFT-B complex. This is consistent with in vitro and in vivo data in Chlamydomonas that showed tight interactions between IFT25 and IFT27 prior to their association to the IFT-B complex (Bhogaraju et al., 2011; Wang et al., 2009).

The flagellum composition of IFT25^RNAi cells was analysed by IFA using, simultaneously, Mab25, an antibody detecting an axoneme-specific protein (Dacheux et al., 2012), and an antibody recognising the IFT-B protein IFT172 (Absalon et al., 2008b). Both antibodies displayed the expected pattern in non-induced cells with the typical axoneme (magenta) and IFT (green) patterns (Fig. S3A). By contrast, staining of IFT25^RNAi cells where RNAi had been induced for 2 days revealed an accumulation of large amounts of IFT172 within the short flagellum (Fig. 4D). The same result was obtained with an antibody against the IFT22 protein (Fig. S4), another component of the IFT-B complex (Adhiambo et al., 2009). This is consistent with a requirement of IFT25 for retrograde transport of IFT-B proteins, as observed for IFT27 (Huet et al., 2014).

The IFT-A complex is required for retrograde transport (Absalon et al., 2008b; Iomini et al., 2009). To monitor IFT-A proteins, the IFT25^RNAi cell line was transfected with a construct allowing expression of a Tandem Tomato (TdT)::IFT140 fusion from its endogenous locus (Huet et al., 2014). In non-induced conditions, TdT::IFT140 trafficked normally within the flagellum (Movie 3; Fig. S3B). By contrast, TdT::IFT140 was immotile in induced IFT25^RNAi cells (Movie 4) and appeared to be stuck at the base of the flagellum with little or no signal within the organelle (Fig. 4E). Therefore, the absence of IFT25 prevents the recycling of the IFT-B complex while it inhibits the access of IFT-A to the flagellum. The absence of IFT-A suggests an uncoupling of IFT-B and IFT-A complexes, which could explain the observed retrograde transport phenotype. This phenotype is quite similar to what was reported upon conditional deletion of IFT25 or IFT27 in mouse male germ cells, where spermatozoa possess much shorter flagella that displayed numerous structural defects (Liu et al., 2017; Zhang et al., 2017). These include aberrant and disorganised axonemes that are too short, defects in extra-axonemal structures, such as the outer dense fibres or the fibrous sheath, and the presence of vesicles in the lumen of the flagellum. Scanning electron microscopy revealed dilation of unknown molecular composition at the distal end, leading the authors to suggest a defect in retrograde IFT (Liu et al., 2017; Zhang et al., 2017).

We propose three hypotheses to explain the different contributions of IFT25 in flagellum assembly. First, the composition of IFT trains could be different between cell types, for example as a consequence of alternative splicing or post-translational modifications. In Chlamydomonas, IFT25 exhibits different phosphorylation patterns and only the phosphorylated version co-sediments with IFT27 on sucrose gradients (Richey and Qin, 2012). If the trains are different, the loss of IFT25 and IFT27 could have variable consequences. Second, IFT25 and IFT27 could be required for the transport of specific cargoes that are expressed in only certain cell types such as members of the hedgehog pathway (Keady et al., 2012; Yang et al., 2015) or the BBSome (Lechtreck et al., 2009). Of note, deletion of BBS genes in mouse does not have a visible impact on primary cilia in the kidney and on motile cilia in the trachea but leads to absence of the sperm flagellum (Mykytyn et al., 2004). If IFT25 and IFT27 contribute to BBS trafficking, the spermatozoa phenotype could be due to BBS perturbation. Third, trypanosome and spermatozoa flagella are characterised by the presence of extra-axonemal structures (Escalier, 2003; Portman and Gull, 2010) that could impact on the displacement of IFT trains. Alternatively, the accumulation of aberrant structures could interfere with proper IFT, and axoneme construction defects would hence be a secondary phenotype. Investigation of the role of IFT25 and IFT27 in different organisms with different types of cilia will be necessary to have a more global vision of their biological role and to understand their loss in so many species.

All procyclic T. brucei cell lines were derivatives of strain 427 and grown in SDM79 medium with hemin and 10% fetal calf serum. The synthesis of the IFT25 coding sequence was conducted by GeneCust Europe (Dudelange, Luxembourg) and the synthesis product was cloned into the pPCPFR vector (Adhiambo et al., 2009) to generate the pPCPFReGFPIFT25 plasmid, allowing GFP tagging at the N-terminal end of IFT25. The resulting plasmid was linearised with NsiI, targeting integration in the intergenic region of PFR2 (Adhiambo et al., 2009). For bimolecular fluorescence complementation assays, constructs were prepared for tagging IFT25 with the first portion of YFP and IFT27 with the second portion. If the two proteins interact closely, YFP should be reconstituted and emit light (Kerppola, 2008). The first 465 nucleotides of YFP were synthesised in a fusion at the 5′ end of the first 468 nucleotides of the IFT25 coding sequence, separated by a linker sequence (encoding GSAGSAAGSG). The resulting construct was cloned in the p2675 vector (Kelly et al., 2007) for in situ insertion in the IFT25 gene upon linearisation with EcoRV and transfection in wild-type cells, resulting in the expression of the YFPNter::IFT25 protein. The final 252 nucleotides of YFP, preceded by an ATG start codon, were synthesised followed by the sequence of the Ty1 tag (Bastin et al., 1996), the linker sequence (encoding GSAGSAAGSG) and the first 440 nucleotides of the IFT27 coding sequence, for expression of the YFP^Cter::IFT27 protein. The synthesised product was cloned in the p2845 vector (Kelly et al., 2007) and linearised with EcoRI to insert in the IFT27 locus transfection in wild-type cells. Screening was carried out by IFA using a polyclonal anti-GFP antibody (cat. no. A6455, Invitrogen) to confirm the presence of each fusion protein. After this validation, the cell line expressing the YFP^Nter::IFT25 protein was next transfected with the construct for expression of the YFP^Cter::IFT27 protein and resistant cells were screened using live fluorescence imaging.

The 29-13 cell line expressing the T7 RNA polymerase and the tetracycline repressor has been described previously (Wirtz et al., 1999). For generation of the IFT25^RNAi cell line, the full IFT25 gene (Tb927.11.13300) consisting of 471 nucleotides was chemically synthesised by GeneCust Europe and cloned in the pZJM vector (Wang et al., 2000), allowing for tetracycline-inducible expression of double-stranded (ds)RNA 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. The plasmid was linearised at the unique NotI site in the rDNA intergenic targeting region before transfection (Wang et al., 2000). To monitor the interactions between IFT25 and IFT27, the pPCPFRGFPIFT25 plasmid was modified to replace the puromycin-resistance gene with a blasticidine-resistance gene to generate the pPCPFRGFPIFT25BLA vector. This plasmid was linearised with NsiI and transformed in the previously described IFT27^RNAi cell line and IFT27^RNAi cell lines expressing the RNAi-resistant versions of IFT27, IFT27T19N and IFT27Q67L (Huet et al., 2014). Trypanosomes were transfected with plasmid constructs by Nucleofector technology (Lonza, Italy) (Burkard et al., 2007). Transfectants were grown in medium with the appropriate antibiotic concentration and clonal populations were obtained by limited dilution.

Cells were fixed and processed exactly as described previously (Huet et al., 2014).

The IFA was carried out as described (Huet et al., 2014). Antibodies used were: mouse anti-IFT27 diluted 1:800 (Huet et al., 2014), mouse anti-IFT22 (1:500) (Adhiambo et al., 2009), mAb25 recognising TbSAXO1 (undiluted), a protein found all along the trypanosome axoneme (Dacheux et al., 2012), mouse monoclonal antibody anti-IFT172 (1:100) (Absalon et al., 2008b) and a commercial rabbit polyclonal anti-GFP (1:500) (Invitrogen). Subclass-specific secondary antibodies coupled to Alexa Fluor 488 and Cy3 (1:400; Jackson ImmunoResearch Laboratories, West Grove, PA) were used for double labelling. Sample observation was performed using a DMI4000 microscope (Leica) and images acquired with an ORCA-03G camera (Hamamatsu, Hamamatsu City, Japan). Pictures were analysed using ImageJ 1.47g13 software (National Institutes of Health, Bethesda, MD) and images were merged and superimposed using Adobe Photoshop CC. For live video microscopy, cells were picked up from the culture, deposited on slides and covered by a coverslip, and observed directly with the DMI4000 microscope. IFT trafficking was recorded with a 250 ms exposure per frame for 30 s using an Evolve 512 EMCCD camera (Photometrics, Tucson, AZ). Kymographs were extracted and analysed as described previously (Buisson et al., 2013; Chenouard et al., 2010).

Samples were prepared and treated exactly as described previously (Huet et al., 2014). For western blots, antibodies used were the anti-IFT27 serum (1:800), the anti-IFT172 (1:500), an anti-GFP antibody (cat. no. 814-460-001, Roche, 1:500) and the monoclonal L13D6 that detects PFR proteins (Kohl et al., 1999) (1:50). For immunoprecipitation, 4 μl of polyclonal rabbit anti-GFP (cat. no. A6455, Invitrogen) or 4 μl of anti-aldolase (kind gift of Paul Michels, Edinburgh, UK) were used in a final volume of 150 μl (Huet et al., 2014).

We thank Monica Bettencourt-Dias and Esben Lorentzen for useful discussions, and Paul Michels and Derrick Robinson for the generous gift of antibodies. We thank the Ultrapole/Ultrastructural Bioimaging of the Institut Pasteur for providing access to their equipment.