A MAP6-Related Protein Is Present in Protozoa and Is Involved in Flagellum Motility

In vertebrates the microtubule-associated proteins MAP6 and MAP6d1 stabilize cold-resistant microtubules. Cilia and flagella have cold-stable microtubules but MAP6 proteins have not been identified in these organelles. Here, we describe TbSAXO as the first MAP6-related protein to be identified in a protozoan, Trypanosoma brucei. Using a heterologous expression system, we show that TbSAXO is a microtubule stabilizing protein. Furthermore we identify the domains of the protein responsible for microtubule binding and stabilizing and show that they share homologies with the microtubule-stabilizing Mn domains of the MAP6 proteins. We demonstrate, in the flagellated parasite, that TbSAXO is an axonemal protein that plays a role in flagellum motility. Lastly we provide evidence that TbSAXO belongs to a group of MAP6-related proteins (SAXO proteins) present only in ciliated or flagellated organisms ranging from protozoa to mammals. We discuss the potential roles of the SAXO proteins in cilia and flagella function.


Introduction
The cytoskeleton of eukaryotic cells is essential to maintain cell structure and polarization, and to optimize membrane dynamics, intracellular transport, cell division, and locomotion. In higher eukaryotes, the cytoskeleton is mainly composed of actin microfilaments, intermediate filaments and microtubules (MTs). MTs are dynamic tubulin polymers where the balance between assembly and disassembly of protofilaments is precisely regulated by microtubule-associated proteins (MAPs) [1]. In vertebrates most MTs will disassemble at low temperature but some remain cold-stable or resistant to drugs such as nocodazole [2,3]. It has been shown that the stability of MTs that are cold-and nocodazole-resistant is largely due to their association with the class of MAP known as MAP6 (previously named STOP for Stable Tubule Only Polypeptide) and MAP6d1 (previously named SL21 for STOP-Like of 21 kD) [2,4,5]. MAP6 and MAP6d1 proteins (designated in this paper as MAP6s) are expressed only in vertebrates [6], and have been localized in neurons, astrocytes, oligodendrocytes, fibroblasts, and several tissues (including heart, muscle, lung, and testis) [4,7,8,9]. MAP6 is expressed as three isoforms (MAP6-1, -2, -3) [2]. These MAPs interact with MTs through their MT binding modules called Mn and Mc: MT resistance to both cold and nocodazole disassembly is via the Mn modules, and cold resistance only is via the Mc modules. Interestingly, the Mn modules are well conserved in MAP6 proteins amongst vertebrates, whereas Mc modules are only found in mammals and are the result of a defined insertion in exon 1 [2,6]. MAP6s modules are bi-functional since MT-stabilizing domains and calmodulin-binding domains overlap to some extent allowing a steric regulation for MT binding as described in vitro and in vivo [4,6]. Additionally, when phosphorylated, MAP6-1 does not bind to microtubules in vitro and co-localizes with actin filaments in vivo suggesting, as for MAP2c (a mammalian neuronal MAP), a role in neurite initiation [10,11,12]. MAP6s share a Nterminal cysteine-rich sequence, which, when palmitoylated, targets these proteins to the Golgi apparatus [2,4]. Map6-null mice exhibit a set of defects similar to those of schizophrenia disorders; thus the loss of Map6 is not lethal but is clearly central for normal synaptic plasticity [13]. Taken together, these data indicate a high level of MAP6s regulation.
Trypanosoma brucei (T. brucei) is a flagellated protozoan parasite that causes human sleeping sickness and Nagana disease in livestock in central and southern Africa. It belongs to the Kinetoplastidae order of protozoa (characterized by the presence of a kinetoplast, the single mitochondrion genome associated with the base of the flagellum), which includes Trypanosoma cruzi and different species of Leishmania that are responsible for Chagas disease and leishmaniasis respectively [14]. The T. brucei cytoskeleton is composed of a subpellicular corset of about 100 MTs underlying the plasma membrane, the mitotic spindle (MS) in mitotic cells, and a single flagellum [15,16]. In the flagellum, the MTs form the canonical axoneme with 9 outer triplets of MTs at the basal bodies level followed by the 9 outer doublets at the transition zone and the 9 outer doublets of MTs surrounding a central pair of MTs (9+2) that is characteristic of motile cilia [17].
The flagellum of trypanosomes participates in a wide variety of functions, from cell mobility to host-parasite interaction (for reviews see [18,19]). It is a highly complex organelle -a proteomic analysis of its cytoskeleton has resulted in the identification of 331 proteins, many of which are conserved in other kinetoplastids and higher eukaryotes [20]. The kinetoplastid flagellum possesses a prominent structure called the paraflagellar rod (PFR), which runs along the axoneme from the point at which the flagellum exits the cell to its distal tip [21] and, which is required for normal flagellum motility [22,23,24,25,26].
In eukaryotes, the MT-based organelles centrioles, cilia and flagella MT have cold-resistant MTs [38,39,40,41,42]. The processes regulating cilia and flagella stability are not fully understood but may involve tubulin post-translational modifications [43,44]. Also, several proteins are involved in their stability such as the ribbon protofilament proteins [45,46,47]. The ribbon protofilament protein Rib43a is conserved in trypanosomes [48] and is found in the cytoskeletal fraction of the flagellum, but has not been studied [20,49,50].
In this study, we present TbSAXO (for T. brucei STOP Axonemal protein) as the first MAP6-related protein identified in a protozoan. We demonstrate that it is an axonemal MAP involved in flagellar motility in T. brucei. In addition, our analysis indicates that TbSAXO belongs to a unique group of MAP6related proteins (SAXO proteins) that are present only in ciliated or flagellated organisms ranging from protozoa to mammals, which suggests for the SAXO proteins unique functions in cilia and flagella of eukaryotes.

Identification of MAP6-related proteins from protozoa to mammals
During a proteomic analysis of T. brucei flagellar proteins [51,52], we identified a 30 kD basic protein (GeneDB accession number Tb927.8.6240). Based on the study (described below), we designated this protein as TbSAXO. TbSAXO is listed in a trypanosome flagella proteome analysis as orthologous to an unknown Plasmodium falciparum protein (PFI0460w), that we designate here PfSAXO [20].
A BLAST analysis, using the TbSAXO sequence, identified orthologues throughout the taxonomic categories and ranged from protozoa to mammals (Fig. S5). Those orthologs were present only in ciliated or flagellated eukaryotes, with the exception of Cryptosporidium, a parasitic non-flagellated, non-ciliated protozoan. Also we were not able to identify orthologues neither in prokaryotes, fungi, nor in the flagellates Naegleria, Phytophthora and Monosiga.
We analyzed the protein primary sequences of TbSAXO, PfSAXO and MmSaxo1 (one of the two orthologs identified in mouse) and identified four shared amino acid features: 1) a high percentage of proline (13.2%, 20% and 11.6% respectively); 2) a cluster of cysteines at the N-terminus (motif 1); 3) a ,15 amino acid repeated domain, which is highly conserved in PfSAXO (motif 2) (Fig. 1A, B); and 4) all three SAXO proteins have a basic calculated pI of 8. 95, 8.25, 9.05 for TbSAXO, PfSAXO and MmSaxo1 respectively [53]. We did not identify any canonical proline-recognition domains in TbSAXO or PfSAXO, but repeated proline-rich sequences in both proteins as well as several proline-recognition domains (SH2 and SH3 binding domains, data not shown) in MmSaxo1 are indicative of a role in proteinprotein interaction [54].
Analysis of the primary sequences of MmMap6s, TbSAXO, PfSAXO and MmSaxo1 indicated that these proteins displayed similar characteristics: 1) a N-terminal domain sequence (motif 1) with a cluster of cysteines (Fig. 1A), and, 2) a repeated domain of 15 amino acids (motif 2) that shared some conserved residues with the MT binding and stabilizing Mn module of MAP6 proteins. The motif 2 is followed by an inter-repeat region (IR) of ,18-20 residues, which is highly conserved in PfSAXO (Fig. 1B). Additionally, we used the motif analysis tool MEME [55,56] with mouse Map6-1, Map6d1, MmSaxo1, PfSAXO, TbSAXO and several other protozoan sequences identified by BLAST to further characterize these proteins, in particular the protozoan proteins (Fig. 1C). Keeping in mind that the highly conserved repeat sequence in PfSAXO introduces a strong bias (Fig. 1B) 1A, B). As shown in Figure 1C, motif 1 is present in the protozoan sequences and the MmSaxo1 sequence but absent in Map6-1 and Map6d1 sequences demonstrating its specificity for a distinct group of proteins (SAXO proteins) despite the common presence of cysteines. Motif 2 is repeated and common to all sequences including the mouse Map6s. This latter motif corresponds partially to the MAP6s Mn modules that are involved in MT resistance to both cold and nocodazole depolymerization that were identified experimentally [6] (Fig. 1B). We thus named the regions containing the motif 2 of TbSAXO, PfSAXO and MmSaxo1 Mn-like motifs (MnL). Based on the highly conserved T/S and F/Y residues respectively at positions 6 and 12 of motif 2, we manually identified additional putative MnL domains (  (Fig. 1B). The C-terminal MnL domains have a longer sequence than the other MnL domains and bare a short Cysteine-Proline motif (CP). We thus differentiate these domains by an asterisk MnL-x* (Fig. 1B, C). Interestingly, the MEME analysis identifies a fourth Mn (or MnL) domain in Map6-1, which has not been identified previously and which experimentally does not seem to be functional [6].
Taken together, these data suggest that the identified SAXO proteins form a family of MAP6-related proteins in ciliated/ flagellated organisms present in protozoa to mammals, and that they are organized as follows: the N-terminus bears a cysteine-rich motif 1 followed by a linear repetition of MnL motifs, the most Cterminal one carrying an additional Cysteine-Proline motif.
TbSAXO is an axoneme specific protein in the protozoan T. brucei During the T. brucei cell cycle, the new flagellum originates from the newly formed basal body and grows through the flagellar pocket (see diagram of a T. brucei cell, Fig. 2A). At the point of exit from the pocket, the flagellum bares the PFR, an accessory structure that is involved in flagellum motility, and is common to some flagellated protists [21]. In the context of the complex flagellum organization and in order to investigate the location and function of TbSAXO in detail, we cloned the gene and produced a polyclonal and a monoclonal antibody (mAb25). Western-blot (WB) analysis on T. brucei whole procyclic form (PCF) cells and bloodstream form (BSF) cells using the polyclonal and mAb25 antibodies showed that the antibodies were specific to TbSAXO and that TbSAXO was expressed in both parasite forms (Fig. 2B). Furthermore, TbSAXO was associated with the cytoskeleton fraction of the flagellum since detergent and salt extraction during the preparation of cytoskeletons and flagella did not remove the protein (Fig. 2C). In addition, WB studies indicated that TbSAXO migrates as a doublet suggesting that one or more post-translational modifications are present on the protein (Fig. 2C).
In both PCF and BSF developmental stages, TbSAXO was observed by immuno-fluorescence (IF) to localize along the flagellum ( Fig. 3 for PCF, Fig. S4 for BSF). Fluorescence signal extended from the transition zone of the axoneme to the distal tip of the flagellum as shown by the co-labeling with the wellestablished transition zone antibody FTZC [57] and the anti-PFR2 antibody L8C4 [58] (Fig. 3A, B). TbSAXO was also present in both the old and new flagella in dividing cells ( Fig. 3B and Fig.  S4). The flagellar signal was visible even when the short, new flagellum was entirely within the flagellar pocket; this was confirmed by the absence of PFR labeling on the new flagellum ( Fig. 3B, 1K1N2F). Double labeling of the PFR and TbSAXO suggested that TbSAXO is an axoneme-associated protein, in that TbSAXO localization extends past the PFR, is proximal to the transition zone, and there was no over-lap or co-localization along the length of the flagellum beside some parts of the flagellum due to the orientation of the cell when observed by immunofluorescence (Fig. 3A, B). Further, immuno-electron labeling of PCF flagella revealed mAb25 labeling all over the axoneme (Fig. 3Ca, b, c). Unfortunately, the stronger fixation conditions required for whole cell preparation prevented immuno-labeling of sectioned material.

TbSAXO is a microtubule-associated and microtubulestabilizing protein
We have shown that TbSAXO is an axoneme-associated protein in the parasite with primary sequence characteristics similar to those of the MAP6s; thus we asked the question could it function in MT binding and/or stabilization?
To study TbSAXO MT stabilizing functions directly, and to circumvent the fact that the trypanosome cytoskeleton is unusually stable [20,27,29,30], we expressed a recombinant TbSAXO-Myc protein in the U-2 OS human cell line (human bone osteosarcoma epithelial cells) [59]. As with many mammalian cell lines, when U-2 OS cells in culture are incubated either at 4uC or in the presence of nocodazole, interphase MTs depolymerize (Fig. S1A, B). To validate the use of the U-2 OS cell system for our purposes, we transfected them with a vector allowing the expression of the rat MAP6-1-GFP protein. After 24 h post-transfection, the cells were briefly detergent-extracted then fixed. At 37uC MAP6-1-GFP was expressed in U-2 OS cells and decorated the interphase MTs (Fig. 4a). After cold or   [4,6]. CP motifs are boxed. IR: inter-repeat regions. Right panel: the Mnlike regular expression is represented as position-specific probability matrix derived from MEME analysis in C. C. Identification of a family of proteins containing Mn-Like domains. MEME analysis using mouse Map6s, mouse Saxo1, and only protozoan SAXO sequences identified a characteristic Nterminal motif (motif 1, dark blue boxes) in SAXO proteins and Mn-like domains (motif 2, light blue boxes) common to the SAXO and MAP6 proteins. Manually identified supplementary motifs 2 are in grey boxes (Motif 2 manual). The asterisk indicates a conserved CP sequence in the last Mn-like domain. doi:10.1371/journal.pone.0031344.g001 nocodazole treatment, only cells expressing MAP6-1-GFP displayed a network of MTs, demonstrating that the MAP6 protein had stabilized the MTs (Fig. 4a, Fig. S1b).
We next transfected the cells with the TbSAXO-Myc expression vector to study its potential to stabilize MTs. At 24 h post transfection, the cells were extracted and fixed and TbSAXO-Myc was detected by immuno-staining using an anti-Myc antibody. At 37uC, TbSAXO-Myc decorated the MT network ( Fig. 4b), similar to what we had observed with MAP6-1-GFP transfection demonstrating that TbSAXO can function as a microtubuleassociated protein in a heterologous system, and can interact with interphase MTs at 37uC. We never observed TbSAXO-Myc labeling on the mitotic spindle or at the midbody, suggesting some specificity for interphase MTs in this system. After cold (4uC) treatment, we observed the depolymerization of MTs in nontransfected cells as would be expected, but in cells expressing TbSAXO-Myc, stabilized MTs were observed to be decorated with TbSAXO-Myc protein (Fig. 4b). In summary, these data show that TbSAXO can act as a MT binding and stabilizing protein.

The MnL motifs are required and sufficient for microtubule binding and stabilization
One of the characteristics of MAP6s is the presence of a Nterminal cysteine cluster, which can be palmitoylated and is responsible for targeting to the Golgi apparatus [4]. We wanted to determine if the N-terminus of TbSAXO was also involved in targeting per se; thus, we first deleted the N-terminal domain In TbSAXO, we identified 6 potential MnL domains (Fig. 1B). We thus investigated the MT-stabilizing properties of the repetitions MnL-1 to MnL-5 (as a block) and the MnL-6* domain (which is longer than the other MnL domains and carries the CP motif) by preparing Myc-tagged constructs and transfecting U-2 OS cells. The C-terminal domain deletion construct (DMnL-6*, TbSAXOD214-266-Myc) did not co-localize with the MT network at 37uC but co-localized with stabilized MTs after 4uC treatment (Fig. 4e). It was thus clear that cold treatment induces a relocalization of the recombinant protein, as previously observed for Map6-3 in NIH3T3 cells [2]. The MnL6* domain (TbSAXO-214-266-Myc) alone was not sufficient for MT targeting and or binding and stabilization since MnL-6* was not associated with the cytoskeleton in either experimental condition (Fig. 4f) although expression of the recombinant protein was confirmed in nonextracted cells (Fig. S2f).
Finally, we deleted the MnL-1 to MnL-5 motif block (TbSAXOD36-213-Myc) and observed that there was no association of this expressed protein with MTs at either 37uC or 4uC ( Fig. 4g and Fig. S2g). Notably, expression of the block of MnL-1 to MnL-5 modules (TbSAXO-36-213-Myc) was able to restore the MT targeting/binding and stabilizing functions after 4uC treatment suggesting a cold-induced re-localization (Fig. 4h). The cold-induced relocalization occurred only in the absence of the MnL6* domain (TbSAXOD214-266-Myc and TbSAXO-36-213-Myc) suggesting a role of this domain in MT-binding at 37uC. Since MnL6* carries a specific Cysteine-Proline motif, we mutated the CP motif to Glycine-Asparagine and showed that the mutation did not affect the MT-binding and stabilizing properties of TbSAXO-Myc (compare b and j in Fig. 4 and Figure S2). This data shows that the CP motif is not involved in MT binding at 37uC and that another signal of unknown nature might be involved.
Full-length or N-terminal deleted TbSAXO-dependent stabilization of MTs was achieved in nocodazole treated cells but required the MnL6 motif since the construct 36-213-Myc did not stabilize microtubules after nocodazole treatment (Fig. S1B). Further, mutation of the CP motif in MnL6* did not modify the MT binding or stabilizing properties of TbSAXO after nocodazole In each case, the transfected cells were incubated at 37uC or 4uC to test for MTs cold stability. Anti-tubulin (TAT1) and anti-Myc antibodies provided the images in left and centre columns respectively at each temperature. The right columns for each temperature set are merged images. The cells were subjected to short extraction before fixation and immuno-labeling. TbSAXO MT stabilization is seen in images b, c, e, h and j. MT stabilization is also observed in the positive control MAP6-1-GFP expressing cells (a). Nuclei were labeled with DAPI. Bar, 20 mm. doi:10.1371/journal.pone.0031344.g004 treatment (Fig. S1Bg), suggesting that this motif is not involved in the nocodazole-stabilizing property of the protein.
Finally, whilst being expressed in U-2 OS cells, the MnL6* domain associated to MnL5 (180-266-Myc) was not able to bind MT neither at 37uC nor at 4uC and did not induce cold or nocodazole MT-stabilization suggesting that more than two MnL domains are necessary (Fig. 4i, S1Bf and S2i).
These data demonstrate that the MnL motifs of TbSAXO do share functional properties with the Mn domains of known MAP6 proteins. MnL6* and MnL5-MnL6* domains are not sufficient for MT targeting/binding or stabilization but modules MnL-1 to MnL-5 are sufficient for MT targeting/binding and stabilization after cold treatment. This suggests that, unlike Map6d1 which has only one Mn module to stabilize MTs, more than two MnL modules might be required for MT binding and stabilization in TbSAXO.
The MnL-1-5 domains are required for flagella targeting/ binding in T. brucei Because of the high stability of trypanosome MTs, the MT stabilization properties of TbSAXO could not be directly assessed in the parasite, but, we investigated the domain(s) involved in flagellum targeting and binding. We transfected procyclic trypanosomes with an over-expression vector allowing the inducible expression of various TbSAXO-Myc truncations of the protein (schematically represented in Fig. 5). For all the transfected cell lines studied, none of the trypanosome (non-induced or induced) developed any growth phenotypes when compared to the non-transfected parental cell line (WT) (data not shown). All Mycconstructs were expressed upon induction as shown by WB using whole cells probed with anti-Myc antibody (Fig. 5) and by IF (Fig.  S3). Immuno-fluorescence experiments on cytoskeleton-extracted cells (Fig. 5) and on whole cells (Fig. S3) showed that, for all constructs, a pool of Myc-tagged protein was found to be soluble, localized in the cytoplasm, and was extracted during the cytoskeleton preparations.
As shown on the WB in Figure 5a, the intensity of the full-length over-expressed protein is only slightly reduced in induced flagella compared to whole cells and cytoskeletons suggesting that most of the recombinant protein is targeted to and associated with the flagellum. After a short induction time of expression of recombinant protein (3-6 h), the tagged TbSAXO was directed to the new but not the old flagellum (Fig. 5a, IF panel) suggesting a slow turnover of the protein or cell cycle dependent targeting. Also, a weak IF signal was observed on the mitotic spindle (MS). This labeling was never observed on WT cytoskeletons probed with mAb25 but was also observed for over-expressed TbSAXO-GFP (unpublished data) suggesting that TbSAXO-Myc is targeted to the newly made flagellum, and that the MS targeting may be an artifact of over-expression. Different expression levels were observed between the truncated forms of recombinant TbSAXO-Myc, as shown on WB (for example Fig. 5b versus  [5c]). Thus, it was not possible to directly compare the flagellar MT affinity of these proteins when analyzing their expression profiles in whole cells, cytoskeletons or flagella.
Our IF and WB data revealed that the MnL-1 to MnL-5 motif block (TbSAXO-36-213-Myc) was necessary and sufficient to target and to bind to the flagellum since its deletion (TbSAXOD36-213-Myc) abolished the axoneme and the MT location of the protein suggesting that the targeting/binding domain has been removed (Fig. 5f, g). Further neither the Nterminal motif 1 (TbSAXO-1-35-Myc) nor the MnL-6* domain (TbSAXO-214-266-Myc) were necessary for flagellum targeting or binding since their tagged version were soluble and their deletion (TbSAXOD2-35-Myc, TbSAXOD214-266-Myc) did not affect the axoneme location of the protein (Fig. 5b-e, Fig. S3). Since in vivo expression of a small recombinant protein might be difficult (TbSAXO-1-35-Myc is 10.2 kD), we expressed a C-terminus GFPtagged version (TbSAXO-1-35-GFP, 30.6 kD). Again, the motif 1-GFP construct was produced as a soluble pool (unpublished data). In contrast to TbSAXOD2-35-Myc and TbSAXOD214-266-Myc, we did not detect any TbSAXO-36-213-Myc labeling at the MS, even after longer induction times when both old and new flagella were decorated (Fig. 5g, IF).
Taken together, these data demonstrate that the MnL domains are required for targeting/binding to the flagellum but that neither the MnL-6* nor motif 1, carrying the cysteine cluster, is required for flagellum targeting or binding. Moreover, these findings support the importance of multiple MnL domains for the MT targeting that we have observed for TbSAXO in mammalian cells (Fig. 4).

TbSAXO is involved in flagellum motility in T. brucei
To assess the functional role of TbSAXO in the parasite, we used the tetracycline-inducible RNA interference (RNAi) system in PCF and BSF cells (kind gifts from G. Cross, Rockefeller University) [57,60,61,62]. Overall, a reduced growth rate was observed for both non-induced PCF and BSF cell lines transformed by the RNAi constructs compared to the nontransformed parental cell line (WT) indicating that there is an induction in the absence of tetracycline or, in other words, a leaky induction of RNAi (Fig. 6a, e). Taking this into account, there was no difference in growth rate observed in induced PCF cells compared to non-induced cells (Fig. 6a). WB analysis of the TbSAXO expression level showed that RNAi induction strongly reduced the expression of TbSAXO in PCF while some reduction of expression was also observed in non-induced cells when compared to WT cells (Fig. 6b). This occurred in numerous clones and after numerous transformations. In BSF cells, induction of RNAi also strongly reduced the expression of TbSAXO (Fig. 6f), some cells retained a very weak TbSAXO IF labeling after 72 h of RNAi knockdown (multinucleate and anucleate cells in Fig. S4B), but this is possibly the consequence of cell death within the 72 h induction period and thus the absence of continued RNAi knockdown.
Both life-cycle stages displayed different RNAi TbSAXO phenotypes but their flagella had consistent length and morphology indicating that flagellar MTs were not destabilized when TbSAXO expression is reduced. TbSAXO RNAi is not lethal in PCF cells, however cell mobility was reduced since cells sedimented suggesting a flagellar motility defect (Fig. 6c). Exemplified by mobility traces (Fig. 6d) and supported by Movie S1, we confirmed the motility defect by illustrating that the RNAi TbSAXO cells remained primarily in one location (,0.4 mm/sec) whilst the WT cells travelled long distances (.10 mm/sec). The RNAi TbSAXO cells were not paralyzed, but the flagellar beat appeared uncoordinated and slower when compared to WT (Movie S2). Also, a small proportion of the PCF cells (5%) were either zoids (1 kinetoplast, no nucleus) or multi-flagellated (4 flagella) indicating some difficulties in cytokinesis when TbSAXO protein level was reduced (Fig. S4A).
Cytokinesis defects have previously been described as a consequence of RNAi knock-down of several proteins involved in flagellar motility in BSF trypanosomes, such as PFR2, TAX-1 and Trypanin, resulting in cells with multiple nuclei and kinetoplasts [20,63]. Even though induction of TbSAXO RNAi in BSF was not unequivocally lethal, non-induced cells displayed a slight reduction in growth rate, which was further accentuated in induced cells (Fig. 6e). Reduction of the expression of TbSAXO in BSF cells (Fig. 6f) induced a cytokinesis-defective phenotype that resulted in an accumulation of cells with aberrant numbers of kinetoplasts, nuclei and flagella ( Fig. 6g and Fig. S4B). Noninduced (leaky) and induced cells initiated multiple S phase and mitosis cycles but appeared to lack subsequent cytokinesis, or in some cases underwent incorrect cytokinesis, leading to accumulation of multinucleated cells and anucleated cells (zoids). The multinucleated and multi-flagellated BSF cells (xKxNxF) had a highly convoluted cell periphery and accumulation of vesicles of unknown composition as observed by electron microscopy on thinsections (Fig. 6h). These aberrant cells are unlikely to recover to the normal phenotype.
To summarize, these data demonstrate that TbSAXO is involved in flagellum motility in PCF and as a consequence in cytokinesis in both forms of T. brucei.

Discussion
In this study, we have identified TbSAXO as an axonemespecific MAP protein in the protozoan T. brucei. By analyzing its primary sequence and performing functional analysis, we have demonstrated that this protein has the characteristics of a structural MAP. Further as discussed in detail below, identification of a cysteine-rich N-terminal domain (motif 1), and of the Mn-like domains (motif 2) involved in stabilization of MTs upon cold and nocodazole treatments, suggests that TbSAXO is the first MAP6related protein to be identified in protozoa. TbSAXO co-localizes with MTs (in both the parasite and mammalian cells), and when expressed in mammalian cells, it stabilizes microtubules and induces MT bundles as is observed for the expression of higher eukaryotic structural MAPs in heterologous cells [64,65,66,67].
Knockdown of TbSAXO using RNAi TbSAXO in its natural context reduced flagellar motility suggesting that TbSAXO plays a role in flagellar motility. The phenotypes induced by the RNAi TbSAXO (flagellar motility defect in PCF and cytokinesis defect in BSF) are not unique; knock-down of other proteins involved in flagellum motility in T. brucei [20,63,68] also produce these phenotypes, even though Ralston et al. demonstrated recently that normal flagellum motility is not required in BSF [69]. Although flagellum motility (in PCF) and cytokinesis (in BSF) are compromised in the RNAi TbSAXO cell lines, no defect in flagellum biogenesis, structure or MT stabilization was observed. This suggests that some TbSAXO that remains after RNAi TbSAXO knockdown is sufficient, or that other protein(s) can compensate for reduced levels of TbSAXO. TbSAXO may have more than one function in trypanosomes. In mammalian cells, we have observed that TbSAXO has intrinsic microtubule stabilization properties as is the case for two other unrelated T. brucei MAPs (CAP15 and CAP17) [36]. This suggests that TbSAXO may also contribute to MT rigidity in the flagellum.
The structural MAPs Tau and Map2 have 3-4 repeats of ,30 amino acids in the microtubule-binding domain, each repeat binding to tubulin. They stabilize microtubules by binding longitudinally individual protofilaments [70,71]. Binding of Tau or Map2c induces an increase of MT rigidity by changing the mechanical properties of the polymer in vitro [72,73]. Our observations suggest a similar mechanism for TbSAXO in flagellar beat or flagellar motility in trypanosome. Our immuno-localization and functional studies suggest that in the wild-type flagellum TbSAXO binds to the protofilaments of outer doublets microtubules, and modulates the rigidity of the polymer thus influencing the beating properties of the flagellum. When TbSAXO level is decreased (by RNAi), the flagellum beating properties may change and are directly reflected by the motility phenotype observed in PCF. From our IF studies, TbSAXO co-localizes with the 9+2 region of the axoneme (distal to the transition zone and continues to the distal tip of the flagellum). The central pair of MTs and associated proteins (also named the central apparatus) is responsible for the motility of the axoneme in motile cilia and flagella [74]. Modulation of the beating properties of the flagellum via TbSAXO is thus coherent with its localization. TbSAXO could therefore bind microtubules to modify or impose rigidity rather than cold resistance. Additionally, the consequences of TbSAXO RNAi may induce compensational modulation of flagellum beat or stability by other proteins, and or may influence the role of the PFR in motility; thus identifying the exact role of TbSAXO cannot be defined without considering the functions of structural and stabilizing proteins and the changes made after TbSAXO knockdown.
In general, MT-based functions are regulated by the recruitment of MAPs via regions carrying specific tubulin posttranslational modifications. In trypanosomes, tubulins are highly post-translationally modified. In particular, acetylated tubulin was identified in the flagellum and subpellicular MTs in T. cruzi [75] and in the flagellum of T. brucei [76]. Also, these MTs are glutamylated and tyrosinated but not glycylated [77,78,79]. By expressing different domains of TbSAXO in mammalian cells and in T. brucei, we have identified the MnL domains as necessary for targeting, binding and stabilizing MTs and for targeting to the flagellum. Since TbSAXO is a MAP, the post-translational status of tubulins might be the ''signal'' or ''code'' required for TbSAXO targeting to the microtubules of the axoneme and not the microtubules of the cellular corset. Although other processes such as intra-flagellar transport (IFT) could also be involved [80], even though IFT has been recently demonstrated not to be specific to ciliated/flagellated cells [81].
TbSAXO and the MAP6 protein share an N-terminal domain carrying a cluster of cysteines, the Mn (or Mn-Like) modules, and cold and nocodazole MT stabilizing property suggesting that TbSAXO is a Map6-related protein. However, the differences in the sequence of the IR regions, the number of repetitions of MnL modules, and the absence of a strict consensus sequence for the Nterminal domain suggest some functional differences. When palmitoylated at the N-terminus, some MAP6s will localize to the Golgi, although their exact function there is not clear [4]. TbSAXO was not observed at the Golgi complex in T. brucei nor U-2 OS cells (our study) and was not identified in a T. brucei protein palmitoylation analysis [82] suggesting a lack of palmitoylation. What could be the role of the N-terminal domain of TbSAXO? When phosphorylated, the structural MAPs such as tau, MAP2, MAP4 and MAP6-1 detach from MTs and can interact with other proteins such as actin [10,83,84,85]. The TbSAXO doublet band observed in our western-blots could reflect phosphorylation of TbSAXO in both PCF and BSF. This is supported by the phosphorylations observed at the N-terminus (serine 8, serine 9 and tyrosine 12) in the phospho-proteome of BSF [86]. There is no direct evidence of a MT-unbound form of TbSAXO in our data but like other MAPs, TbSAXO could have a modified MT-binding profile or structure in the phosphorylated state thus modulating the rigidity of the MTs in response to extracellular signals. Additionally, the presence of several prolinerecognition domains in TbSAXO could suggest interactions with other flagellar proteins [54].
Until now MAP6 proteins have only been identified in vertebrate cells. Axoneme localization and a role in flagellum motility have not been described to date for any of the MAP6 proteins. Our databases/literature searches identified SAXO orthologs from protozoan to mammals but only in ciliated/ flagellated organisms, with the exception of Cryptosporidium. A specific role for SAXO proteins in cilia and flagella is strongly suggested since proteomes of flagella of sperm cells (in rat, mouse and drosophila), flagella of T. brucei, Chlamydomonas and cilia of human airway cells have SAXO orthologs [20,87,88,89,90,91]. A SAXO ortholog was identified in a total extract of C. elegans, which has ciliated sensory neurons [92]. In contrast, no SAXO orthologs was listed in the mouse photoreceptor neuron proteome. Map6 does appear in samples containing the cytoskeleton rootlet but not in the sample containing only the axoneme [93]. Taken together this points to a specific role for SAXO proteins in flagella and cilia function. We predict that data generated from additional analyses would clearly open new avenues in the study of this protein family, and its role in axonemal function and diseases.

Ethics statement
All animal experiments were performed in accordance with institutional guidelines as determined by the Service Commun des Animaleries de l'Université Bordeaux Segalen (approval number A33-063-916), which specifically approved the study. All experiments requiring genetically modified organisms were carried out with approval of the ethics committee of the Ministère de l'Enseignement Supérieur et de la Recherche (approval number 4946).
In silico analysis SAXO orthologues were identified by walking BLASTP (BLAST 2.2.26 [94,95,96]) using default parameters (non redundant protein sequences database, BLOSUM 62 matrix, gap cots existence: 11; extension: 1; conditional compositional score matrix adjustment, no filter, no mask) and a cut-off E-value of 4e-06 using initially TbSAXO or PfSAXO as query. Identified proteins were then also used as query proteins.
Settings for MEME analysis [55] were: any number of repetition, optimum width of each motif of 6 minimum and 30 maximum, 2 maximum motifs. Accession numbers of protein sequences used for the MEME analysis are: Mus musculus Map6-1 (NP_034967.

Statistical analysis
In all experiments, error bars represent standard deviation (SE) of samples in triplicates. Statistics in Figure 6g were performed using an unpaired two-tailed t-test with a 90%, 95% or 99% confidence interval. XLSTAT software (XLSTAT.com) was used to determine P-values.

Cell lines, growth conditions and transfection
Genomic DNA of the T. brucei cell line TREU927/4 GUTat10.1 (a kind gift from S. Melville, Cambridge University) selected by the genome project [97] was used to amplify by PCR the ORFs of the proteins studied here. The work described in this study uses the parental PCF and BSF T. brucei 427 strains (kind gifts from G. Cross, Rockefeller University), co-expressing the T7 RNA polymerase and a tetracycline repressor, named in this study as wild-type (WT) [98]. Over-expression and RNAi were induced with 1 mg.mL 21 tetracycline. PCF trypanosomes (T. brucei 427 29-13) were grown and transfected as in [99]. BSF trypanosomes (T. brucei 427 90-13) were grown and transfected as in [100]. Special care was taken to not keep in culture the selected PCF and BSF transformants more than 2 weeks because of revertion over longterm culture [101]. Transformants were screened by immunofluorescence, after tetracycline induction and cloned by serial dilution. Correct integration in the PCF mini chromosomes (MCs) of the pPRP-177 vector was checked by pulsed-field electrophoresis and southern-blot as described in [62,102]. Growth curves were done by using a mallassez cell counter every 24 h and by diluting the PCF cells back to 2.10 6 cells/ml and the BSF back to 3.10 5 cell/ml. Growth curves in Figure 6 represent the cumulative cell number.

Plasmid construction
The TbSAXO ORF was amplified from T. brucei 927 genomic DNA using the two specific primers EcoRI-1019 (59-CCCGAATTCATGACAACATTGCACACTATTTCCAGCC-39) and 1019-HindIII (59-AAAAAAGCTTCTAATCCGCAA-TTTCTCCTGAGGGG-39) and cloned into pET28a+ (Novagen) in frame with a N-terminal 6-histidine-tag making the pDRET1 plasmid used for bacterial expression and purification. To produce the TbSAXO RNAi cell lines (PCF and BSF), a 650 bp insert was amplified by PCR (primers 600PRP-XbaIBamHI 59-TTCA-GAGGGATCCCTCGAGTCTAGAAAGTCACGAACGTCG-GGCATCGATT-39 and HindIII-1019 59-AAAAGTTCGAA-GCTTATGACAACATTGCACACTATTTCCAGCC-39) and cloned into the p2T7-177 plasmid [62,102] thus providing pPRP-177. We have checked the pPRP-177 vector by sequencing the 177 pb repeat sequences, the promoter regions and the RNAi target sequence. The NotI linearized plasmid was transfected in PCF and BSF cells. For the expression of Myc-tagged TbSAXO, the full-length TbSAXO ORF lacking the stop codon was cloned into the HindIII-XbaI sites of pLew100X-Myc vector containing a C-terminal 36Myc-tag and of a pLew100X-GFP plasmid containing a C-terminal GFP (pLew100 vector from [98]  Production of the anti-TbSAXO monoclonal antibody 100 mg of 6His-TbSAXO was emulsified with complete Freund's adjuvant (for first injection) or incomplete Freund's adjuvant (following boosts) and used to immunize two Balb/C female mice (Charles River Laboratories, L'Arbresle, France). Spleen cells of one mouse were fused to myeloma X63-Ag8 cells as in [104]. One hybridoma producing a TbSAXO-specific antibody was selected and named mAb25 (IgG2a).

T. brucei flagellar protein preparation
Flagellar proteins were prepared, separated in denaturing conditions on a BIORAD ROTOFOR (ampholins 3-10) followed by SDS-PAGE electrophoresis as in [52]. After Coomassie staining, two protein bands were excised, trypsin-digested and subjected to LC-MS/MS.

Sedimentation assays and video-microscopy
Sedimentation assays were done as in [23]. Briefly, WT, noninduced (2tetracycline) and induced (+tetracycline) RNAi TbSAXO cells were grown for 24 h then placed in cuvettes. The OD 600 nm was measured at t = 0 and t = 24 h before (ODb) and after (ODa) mixing. The graph represents the percentage of sedimentation calculated as 100-(ODb/ODa 6100). Video-microscopy was carried out as described in [106].
U-2 OS cells. For the MT stability assays in U-2 OS cells at 37uC and 4uC (cells were incubated 30 min at 4uC), all solutions were pre-warmed at 37uC or cooled at 4uC, including the fixation step; alternatively cells were incubated in medium supplemented with 20 mM nocodazole 30 min, 37uC, to induce MT depolymerization. For observation of whole cells, cells grown on coverslips were washed briefly with 30 ml of EMT buffer (PIPES 60 mM, HEPES 25 mM, EGTA 10 mM, MgCl 2 10 mM adjusted to pH 6.9 with KOH) and fixed in paraformaldehyde 3% in EMT for 15 minutes (at 37uC or 4uC). To remove soluble proteins (including non-MT associated proteins), cells were briefly extracted for 2 min with 30 ml of EMT, TX-100 0.5%, glycerol 10% then fixed in paraformaldehyde (at 37uC or 4uC, 15 min). Cells were neutralized 10 min in glycine (100 mM in PBS). After two washes in PBS, cells were incubated in PB (PBS, 10% fetal calf serum, 0.01% saponin) for 10 minutes. Primary antibodies anti-cMyc (rabbit polyclonal Sigma c-3956, 1:100) and anti-tubulin (TAT1, 1:100, [107]) were added for 1 hour in a dark moist chamber. After two PBS washes, cells were incubated for 1 hour with the secondary antibodies: anti-rabbit FITC conjugated (sigma F-9887, 1:400) and anti-mouse IgG (H+L) Alexafluor-647 conjugated (Molecular probes, A-21235, 1:400) respectively. The nuclei were stained with DAPI (1 mg.mL 21 in PBS for 5 minutes) and cells were washed and mounted as described above.
Images were acquired on a Zeiss Imager Z1 microscope, using a Photometrics Coolsnap HQ2 camera, at 1006 or 636 (NA 1.4) with MetamorphH software (Molecular Devices), and processed with ImageJ.

Electron microscopy and immuno-electron microscopy
Cells were prepared as in [99] with the following changes: 50 mL of log phase BSF WT and RNAi TbSAXO cells (72 hours induction) were used. Cells were fixed in 25 mL of 4% glutaraldehyde, 4% paraformaldehyde, and 0.5% tannic acid, in 0.1 M cacodylate buffer pH 7.4. Immuno-labeling of flagella was done as follows. Flagella were prepared as in [52] and resuspended at 1.7610 7 flagella/mL. Ten mL of purified flagella were mixed with 100 mL of PBS and place on a sheet of clean Parafilm. Charged Butvar and carbon-coated nickel grids (EMS G200-Ni) were floated onto the droplet for 15 minutes. The flagella were fixed by transferring the grids to 500 ml of 2% Paraformaldehyde, 0.025% Glutaraldehyde in PBS for 30 min. Grids were then neutralized by transferring to 500 ml droplets of 100 mM Glycine in PBS 2610 min. Grids were transferred to 30 mL of mAb25 (neat) for 4 hours. The grids were washed (three times 10 minutes) in PBS and then incubated on 30 mL drop of 10 nm gold conjugate protein A (Aurion) (diluted 1:10 in PBS, overnight at 4uC). Grids were then washed 4610 min in PBS, then fixed 30 minutes in 100 mL droplets of 2.5% glutaraldehyde in distilled water, washed 465 min on 500 mL droplets of water, and negatively stained (NanoVan, Nanoprobes; 40 mL grid). Figure S1 The U-2 OS cell line is a valid model for the cold-and nocodazole-induced MT depolymerization test. The MTs of U-2 OS cells were labeled with anti-tubulin and nuclei with DAPI. A. IF on mock transfected U2-OS cells subjected to 37uC/4uC treatment and a short extraction before fixation. B. MAP6-1-GFP (b), TbSAXO-Myc (c) or various truncated versions (d-f) and a mutated version of TbSAXO-Myc (g) was expressed in U-2 OS cells and tested by IF after nocodazole treatment. In each panel, the last columns are merged images. Scale bars represent 20 mm. (TIF) Figure S2 IF on U-2 OS whole cells expressing TbSAXO-Myc truncations or MAP6-1-GFP. Experimental conditions were as in Figure 4 except that the cells were fixed before permeabilization in order to visualize the soluble pool of the recombinant proteins and to demonstrate their expression. In each temperature regime, the left column shows tubulin (red), the center column shows the recombinant protein (green), and the right column the merged images. Scale bar represents 20 mm. A. MEME analysis, including SAXO orthologues from protozoa to mammals but also MAP6 proteins from different vertebrates, identified the motif 1 (in red), motif 2a (dark blue), and motif 2b (light blue). B. Position-specific probability matrix derived from the MEME analysis for motif 1, motif 2a and motif 2b and their respective regular expression. In this widerange analysis, the initial motif 2 identified in Figure 1B