The tubulin cofactor C family member TBCCD1 orchestrates cytoskeletal filament formation

Tubulin cofactor C (TBCC) interacts with other tubulin cofactors to stimulate the GTPase activity of β-tubulin to release α-tubulin–β-tubulin heterodimers competent for microtubule incorporation (Tian et al., 1996). Human XRP2 (a TBCC-domain containing protein mutated in certain forms of retinitis pigmentosa – a retinal degenerative disease) is proposed to act as GTPase-activating protein (GAP) for protein(s) other than tubulin, albeit in a cytoskeletal context (Veltel et al., 2008; Schwarz et al., 2012). In canonical TBCC and XRP2, stimulation of GTPase activity is dependent upon a catalytic arginine residue (Bartolini et al., 2002). TBCCD1 is an enigmatic TBCC family member. In the green alga Chlamydomonas reinhardtii, TBCCD1 is found associated with the centrioles and rhizoplasts (fibres that connect centrioles to the nucleus; Brugerolle and Mignot, 2003). In Chlamydomonas, centrioles are multifunctional and are involved in both axoneme formation and mitotic spindle organisation. In the absence of TBCCD1 defects in centriole number, centriole positioning and mitotic spindle organisation are seen (Feldman and Marshall, 2009). In animal cells, TBCCD1 is required for centrosome–Golgi linkage (Gonçalves et al., 2010), but in both Chlamydomonas and animal cells how TBCCD1 serves its function is unknown. At first glance, the presence of a TBCC domain suggests that GAP activity is intrinsic to TBCCD1 function. However, TBCCD1 lacks the obvious catalytic arginine of TBCC and XRP2, although in their alignment of TBCCD1, XRP2 and TBCC orthologues Feldman and Marshall (Feldman and Marshall, 2009) identified an arginine residue nearby. The authors speculated this arginine could suffice for GAP activity and suggested that TBCCD1 might function in processing tubulin, including atypical centriole-associated δ- and ε-tubulin. However, other amino residues within the TBCC domain that are conserved in TBCC and RP2, and that are probably crucial for catalytic function (Kühnel et al., 2006), are not conserved in TBCCD1 despite a prediction of conserved secondary structure (our unpublished observations; see Materials and Methods). Moreover, whereas TBCC and XRP2 partially complement the yeast CIN2 (TBCC) deletion mutant (Bartolini et al., 2002), TBCCD1 cannot (Gonçalves et al., 2010). Thus the role of TBCCD1 as a GAP is questionable. Trypanosoma brucei is the aetiological agent of African sleeping sickness, but, from a cell biology perspective, cell form is governed by an elaborate microtubule-based cytoskeleton in which centrioles (or basal bodies) are the template for axoneme formation (Portman and Gull, 2012) but mitosis is closed and occurs independently of centriole involvement (Ogbadoyi et al., 2000). Here, we reveal trypanosome TBCCD1 is an essential protein that functions in diverse aspects of filament-based cell biology with no evident role in microtubule organisation.

To localise T. brucei TBCCD1 (TbTBCCD1; encoded by Tb927.11.2440 and identified as a TBCCD1 orthologue in Feldman and Marshall, 2009), we expressed the protein fused to the C-terminus of yellow fluorescent protein (^TyYFP:TBCCD1). Using a pENT-based vector system (Kelly et al., 2007), recombinant gene expression occurred from an endogenous chromosomal locus under the control of native cis-acting elements. Visualisation of ^TyYFP:TBCCD1 at early stages of the cell division cycle (1K1N) revealed several foci of YFP fluorescence: (1) close to the kinetoplast (an elaborate concatenation of ∼5000 DNA circles forming the trypanosome mitochondrial genome; Jensen and Englund, 2012), (2) at an elongate structure immediately anterior to the kinetoplast, and (3) at the anterior end of the cell body (Fig. 1A). This pattern was retained in detergent-extracted cytoskeletons, indicating tight cytoskeletal association. To look further at localisation and determine the temporal order of signal duplication, we stained detergent-extracted cytoskeletons for indirect immunofluorescence using antibodies recognising various cytoskeletal structures (Fig. 1B–I). Monoclonal antibody YL1/2 recognises a pool of tyrosinated α-tubulin at transitional fibres located at the distal end of the mature basal body (Stephan et al., 2007) and monoclonal antibody BBA4 recognises an antigen from the proximal ends of both mature and pro-basal bodies (Woods et al., 1989). In dual-fluorescence experiments, two ^TyYFP:TBCCD1 foci were positioned more proximally to the kinetoplast than YL1/2, but distal to antigen recognised by BBA4 (Fig. 1B,C). Thus, one pool of ^TyYFP:TBCCD1 coincides with the microtubule barrels of mature and pro-basal bodies.

The T. brucei flagellum is attached to the cell body for most of its length via the flagellum attachment zone (FAZ). Monoclonal antibody L3B2 recognises FAZ1 protein from the cytoplasmic face of the FAZ (Vaughan et al., 2008). In trypanosomes with one flagellum, and which are therefore at either the start or the end of their cell division cycle, L3B2 immunofluorescence revealed ^TyYFP:TBCCD1 immediately adjacent to the distal end of the FAZ signal, consistent with recruitment of ^TyYFP:TBCCD1 to the anterior end of the cell body (Fig. 1D). Careful analysis of cells at various stages of the cell division cycle revealed the new ‘anterior end’ signal did not track with the new FAZ that elongates during new flagellum assembly. Rather, new signal appeared during mitosis; duplication occurred early in mitosis and the duplicated signal (Fig. 1E, arrowed) was acquired prior to full FAZ maturation (as defined by L3B2 labelling). At the anterior end of trypanosome cells, the minus ends of microtubules forming the sub-pellicular corset are clustered (Sherwin and Gull, 1989); duplication of the anterior end ^TyYFP:TBCCD1 signal potentially marks the site for cleavage furrow ingression, which initiates at this point during cytokinesis (Fig. 1F).

Position and appearance of the elongate ^TyYFP:TBCCD1 signal adjacent to basal bodies was immediately reminiscent of the trypanosome-specific filamentous ‘bi-lobe’ (Esson et al., 2012) implicated in providing a template for cell-cycle-regulated Golgi biogenesis and duplication in T. brucei (Ho et al., 2006). Fluorescence and immunoelectron microscopy of known bi-lobe proteins, including centrin isoforms and various trypanosomatid-specific proteins, reveal the bi-lobe to possess a complex sub-structure (Esson et al., 2012; Gheiratmand et al., 2013; Morriswood et al., 2013). Immunofluorescence with either the monoclonal pan-centrin antibody 20H5 (Fig. 1G) or cyan fluorescent protein (CFP)-tagged to the bi-lobe protein TbMORN (CFP:MORN) (Fig. 1H) confirmed TbTBCCD1 as a novel bi-lobe component. However, TbTBCCD1 did not localise fully to either of two known bi-lobe sub-structures: the ‘fishhook’ denoted by TbMORN or the centrin-containing bi-lobe arm as described by Esson et al. and Morriswood et al. (Esson et al., 2102; Morriswood et al., 2013). Immunofluorescence with polyclonal antiserum recognising the Golgi marker RabX2 (Natesan et al., 2009) confirmed the presence of the Golgi stack adjacent to the TbTBCCD1-containing bi-lobe (Fig. 1I). Expression of TbTBCCD1 fused at its C-terminus to GFP (TBCCD1:GFP^Ty) gave an identical pattern of fluorescence to ^TyYFP:TBCCD1. Signal amplification using anti-YFP antibodies failed to detect further ^TyYFP:TBCCD1 foci at any point in the cell cycle. Crucially, given the lethal RNAi phenotype described below, cells expressing only TBCCD1:GFP^Ty or ^TyYFP:TBCCD1 grew without discernible morphological defects. The illustrations in Fig. 2 summarise the localisation of TbTBCCD1 to three distinct sites within trypanosomes, and also indicate the timing of signal duplications during the cell division cycle.

To probe TbTBCCD1 function we induced RNA interference (RNAi) in cells expressing ^TyYFP:TBCCD1. Following RNAi induction, retardation of the growth rate and reduction in TbTBCCD1 protein levels occurred (Fig. 3A,B). No perturbation of microtubule organisation was evident by either light or electron microscopy. In contrast, CFP:MORN localisation (Fig. 3D,E) and immunofluorescence using 20H5 (data not shown) revealed loss of normal ‘bi-lobe’ architecture, indicating TbTBCCD1 plays an important role in organising the formation of this structure. A flagellum detachment phenotype in some RNAi-induced cells suggested either (1) synthesis and vesicular trafficking via the Golgi of surface-associated glycoconjugates was potentially affected by loss of bi-lobe organisation – stable attachment of the flagellum to the cell body is dependent upon members of the trypanosomatid-specific FLA1 glycoprotein family (discussed further in Sun et al., 2013; Ginger et al., 2013) – or (2) the flagellar neck region, as well as bi-lobe organisation, was disrupted following RNAi induction, such that flagellum–cell-body attachment at the flagellar pocket exit point could become compromised.

More striking than losing bi-lobe organisation was the accumulation, post-RNAi induction, of cells either lacking kinetoplasts or possessing abnormally large kinetoplasts (Fig. 3C,G). The kinetoplast is attached to mature and pro-basal bodies by a complex filament-based tripartite attachment complex (TAC) that traverses inner and outer mitochondrial membranes (Ogbadoyi et al., 2003; Jensen and Englund, 2012). Filaments connecting basal body and pro-basal body with the mitochondrial outer membrane are termed exclusion zone filaments, as their presence excludes cytoplasmic ribosomes (Ogbadoyi et al., 2003). The TAC couples basal body segregation with replicated kinetoplast segregation, providing faithful inheritance of single-unit mitochondrial genomes at cytokinesis (Robinson and Gull, 1991). Following RNAi induction, in cells with segregated basal bodies but only one kinetoplast, we observed it was the more posterior of the two basal bodies (i.e. the basal body from which the new axoneme extended) that associated with either little or no kinetoplast DNA (kDNA). In these cells, the more anterior basal body was linked to a larger than normal, fluorescence-intense kinetoplast (Fig. 3G); the size and fluorescence intensity of the large kinetoplasts indicated kDNA replication still occurred. Similar increases in kinetoplast size, corresponding to continued kDNA replication were observed in RNAi mutants deficient in mitochondrial TAC components (Zhao et al., 2008; Sykes and Hajduk, 2013). Our observations indicate loss of centriolar TbTBCCD1 causes asymmetric mis-segregation of kDNA.

Given that centriolar TbTBCCD1 lies more distal than antigen recognised by BBA4 (Fig. 1C), TbTBCCD1 would appear unlikely to be an integral component of extra-mitochondrial TAC filaments forming the zone of ribosome exclusion between the proximal end of basal bodies and the underlying mitochondrial membrane (Fig. 4A,B). When RNAi mutants were examined by transmission electron microscopy (TEM), retention of a ribosome exclusion zone beneath akinetoplastic basal bodies was evident (Fig. 4C,D). We suggest that TbTBCCD1 could orchestrate linkage of extra-mitochondrial TAC filaments to mature and pro-basal bodies, thereby mediating a link between the filament-based TAC and the microtubule barrels of pro- and mature basal bodies. Following TbTBCCD1 RNAi induction and an initial round of kinetoplast replication, the posterior mature/pro-basal body pair segregate without attached kDNA. If TbTBCCD1 does link extra-mitochondrial TAC filaments to basal bodies, then retention of a ribosome exclusion zone beneath akinetoplastic basal bodies (Fig. 4C,D) can be explained if TAC nucleation occurs on mitochondrial membranes and proceeds bi-directionally towards the kinetoplast and basal bodies. Alternatively, ultrastructural analyses of diverse trypanosomatids suggest that the TAC architecture is fundamentally conserved throughout the trypanosomatid family (Ogbadoyi et al., 2003; Jensen and Englund, 2012; Souto-Padrón et al., 1984). From the assorted fixations that have informed on TAC architecture, filaments (or fibrils) directly attaching basal bodies to kinetoplasts have been noted in freeze-etch micrographs of the American trypanosome T. cruzi (Souto- Padrón et al., 1984). Thus, TbTBCCD1 could be a part of, or orchestrate the formation of, these filaments. Given the complexity of TAC architecture, loss of these filaments might not necessarily prevent the formation of exclusion zone filaments that serve to link basal bodies to the outer mitochondrial membrane, thereby providing another explanation for the retention of exclusion zone filaments beneath akinetoplastic basal bodies.

Detailed analyses of TAC–kinetoplast association are currently limited to mitochondrial proteins (Ochsenreiter et al., 2008; Zhao et al., 2008; Sykes and Hajduk, 2013). We therefore looked further, by using TEM, at the kDNA mis-segregation phenotype arising following RNAi against TbTBCCD1 (a non-mitochondrial, TAC-associated protein). In thin sections from non-induced cultures, basal bodies were always found in association with a kinetoplast (Fig. 4A,B). In contrast, thin sections at 48 hours post-RNAi induction revealed basal bodies associated with no kinetoplast (Fig. 4C,D), or unusually long kinetoplasts with the basal body association to that kinetoplast being generally asymmetric (Fig. 4E). In some cells with unusually long kinetoplasts, ordered stacks of kDNA were observed, specifically in the region lacking an overlying basal body (Fig. 4E,F). The kDNA comprises of two classes of circular DNA: maxi-circles (n = ∼40) containing protein-coding genes and mini-circles (n = ∼5000) encoding guide RNAs required for decrypting maxi-circle-derived transcripts. Current kDNA replication models propose that mini-circles move from the kDNA disc into a region situated between the kDNA network and overlying basal body (the kinetoflagellar zone, KFZ; reviewed in Jensen and Englund, 2012). Mini-circle replication occurs within the KFZ, before sister mini-circles migrate to antipodal sites for reattachment to the kDNA periphery. Intriguingly, the presence of electron-lucent regions between well-defined kDNA stacks suggests that despite kinetoplast mis-segregation a structural framework assembles on which replicated kDNA circles then catenate. The electron-lucent regions are potentially analogous to the filament-based KFZ within which mini-circle replication occurs.

Feldman and Marshall (Feldman and Marshall, 2009) noted TBCCD1 was required for mother–daughter centriole linkage in Chlamydomonas. Thus, we asked what effect TbTBCCD1 RNAi had on basal body linkage in trypanosomes. Using BBA4 as an immunofluorescence marker, we saw no evidence for widespread disengagement of pro- and mature basal bodies. Accumulation of anucleate zoids in RNAi-induced populations (Fig. 3C) indicates defects in nuclear positioning following mitosis occur readily. Any filament-based network responsible for nuclear segregation awaits characterisation.

The evidence that TBCCD1 is a GAP for either tubulin or another protein is equivocal – the bioinformatic comparison of TBCCD1 orthologues with TBCC and RP2 does not highlight strong conservation of residues linked to GAP activity in XRP2 (our data not shown). Moreover, a variety of experiments reported by Gonçalves et al. (Gonçalves et al., 2010) indicated no direct involvement for human TBCCD1 in microtubule organisation. Gonçalves et al. (Gonçalves et al., 2010) speculated that TBCCD1 could promote crosstalk between the centrosome and actin-based cytoskeleton in mammalian cells. Our data indicate TBCCD1 is involved in filament formation, structures notoriously difficult to visualise by conventional EM approaches. TBCCD1 presence in Chlamydomonas rhizoplasts, striated fibres connecting proximal ends of flagellar basal bodies to the nucleus, is consistent with a filament function in this protist, too (and loss of mother–daughter centriole linkage explained by loss of another TBCCD1-dependent filament-based structure). Moreover, TBCCD1–Golgi association is conserved between animals and trypanosomes and probably algae, too (given that rhizoplast–Golgi association has been noted; Brugerolle and Mignot, 2003). The last common ancestor of green algae, trypanosomes and animals was probably the last common ancestor of all eukaryotes (Katz, 2012). A TBCCD1-dependent Golgi-to-centriole or Golgi-to-centrosome linkage has potentially therefore been conserved throughout eukaryotic evolution. Coupled with our unequivocal connection of TbTBCCD1 to two filament-based cytoskeletal structures, it is possible that collectively the available data unexpectedly point to the essential co-option of a key domain required classically in tubulin-folding for a non-tubulin, non-GAP-related function at an early stage in the evolution of the eukaryotic cytoskeleton.

Trypanosomes (SmOxP927; Poon et al., 2012) were cultured in SDM-79 medium supplemented with 10% (v/v) heat-inactivated fetal bovine serum (Brun and Schönenberger, 1979). Stable transformation used standard methods.

Fluorescent fusion proteins were expressed using pEnT-based vectors (Kelly et al., 2007). For RNAi, a stem-loop construct (bp 22–1158 of Tb927.11.2440 coding sequence ligated as an inverted repeat to bp 754–25) was cloned between opposing head-to-head T7 RNA polymerase promoters in p2T7₁₇₇ (Wickstead et al., 2002). Primer sequences are available on request. Plasmid sequencing used ABI prism sequencing technology.

Cells were settled onto coverslips and either fixed directly with 3.7% paraformaldehyde or detergent-extracted for 45 seconds with 1% Nonidet-P40 in 0.1M PIPES pH 6.9, 2 mM EGTA, 1 mM MgSO₄ and 0.1 mM EDTA, prior to fixation. Processing continued with a 10-minute incubation in −20°C methanol, followed by rehydration in phosphate-buffered saline (PBS). TbRP2-specific polyclonal antiserum (prepared by Eurogentech and affinity-purified) was used at 1∶50 dilution with a 1% BSA block (immunofluorescence) or 1∶100 with 1% Marvel block (immunoblotting). Immunofluorescence with monoclonal antibodies YL1/2 (Kilmartin et al., 1982), BBA4 (Woodward et al., 1995), L3B2 (Vaughan et al., 2008), RabX2 (Natesan et al., 2009) and 20H5 (Millipore) was performed as described previously or following supplier's instructions (20H5). Images were captured and processed using an Applied Precision DeltaVision Deconvolution microscope system and SoftWoRx software. Nuclei and kinetoplast counts were determined using a Leica DM RXA2 microscope and associated FW4000 software. Immunoblotting using BB2 was as described previously (Bastin et al., 1996).

Fixation was by addition of glutaraldehyde (2.5% final concentration, 5 minutes) to cultures. Cell pellets were re-suspended in 0.1M PBS (pH 7.4) for 10 minutes, followed by 2.5% glutaraldehyde, 2% paraformaldehyde and 0.1% tannic acid in 0.1M phosphate buffer (pH 7.0) for 2 hours at room temperature. Pellets were washed with 0.1 M phosphate buffer (pH 7.0) and post-fixed in 1% osmium tetroxide in 0.1M phosphate buffer (pH 7.0) for 1 hour at room temperature. Samples were rinsed and stained en bloc for 40 minutes in 2% uranyl acetate, dehydrated in an ascending acetone series and embedded in Agar 100 resin (Agar Scientific). Random thin-section electron microscopy was performed using a Hitachi H-7650, operated at 120 kV.

Protein sequences were aligned by Clustal Omega (Sievers et al., 2011) and secondary structure predictions were carried out using Phyre² (Kelley and Sternberg, 2009).

We thank Keith Gull (University of Oxford, UK) for kindly providing the monoclonal antibodies BBA4 and L3B2 and Mark Field (University of Cambridge, UK) for the polyclonal antiserum recognising RabX2.