ABSTRACT
Trypanosoma brucei, the causative agent of African sleeping sickness, has a flagellum that is crucial for motility, pathogenicity, and viability. In most eukaryotes, the intraflagellar transport (IFT) machinery drives flagellum biogenesis, and anterograde IFT requires kinesin-2 motor proteins. In this study, we investigated the function of the two T. brucei kinesin-2 proteins, TbKin2a and TbKin2b, in bloodstream form trypanosomes. We found that, compared to kinesin-2 proteins across other phyla, TbKin2a and TbKin2b show greater variation in neck, stalk and tail domain sequences. Both kinesins contributed additively to flagellar lengthening. Silencing TbKin2a inhibited cell proliferation, cytokinesis and motility, whereas silencing TbKin2b did not. TbKin2a was localized on the flagellum and colocalized with IFT components near the basal body, consistent with it performing a role in IFT. TbKin2a was also detected on the flagellar attachment zone, a specialized structure that connects the flagellum to the cell body. Our results indicate that kinesin-2 proteins in trypanosomes play conserved roles in flagellar biosynthesis and exhibit a specialized localization, emphasizing the evolutionary flexibility of motor protein function in an organism with a large complement of kinesins.
INTRODUCTION
Trypanosoma brucei spp. are kinetoplastid parasites that cause African trypanosomiasis (African sleeping sickness) in humans and animals (Büscher et al., 2017; Cayla et al., 2019; Kennedy and Rodgers, 2019). They are transmitted to mammals by the tsetse fly, and progress through multiple life-cycle stages, including the insect vector procyclic form (PCF) and infective mammalian bloodstream form (BSF). In human African trypanosomiasis (HAT), the BSF proliferates extracellularly in the blood and then in the central nervous system (Kennedy and Rodgers, 2019; Krüger et al., 2018; Mogk et al., 2017). Recent progress in treating HAT, particularly the development of the oral agent fexinidazole (Mesu et al., 2018), is encouraging. Nonetheless, HAT remains a significant public health threat in sub-Saharan Africa (Kennedy, 2019). Moreover, most current treatment options have significant issues with drug resistance, toxicity, cost and difficulty of administration (Akinsolu et al., 2019; Fairlamb and Horn, 2018; Field et al., 2017). Thus, the development of new drugs remains critical.
T. brucei is highly motile, and its motility is driven by a single membrane-bound cilium (flagellum) containing a canonical 9+2 microtubule axoneme with a filamentous paraflagellar rod (PFR) (Höög et al., 2012; Langousis and Hill, 2014; Ralston et al., 2009). The flagellum attaches to the cell body along the flagellar attachment zone (FAZ), a specialized structure running beneath the cytoplasmic face of the plasma membrane along most of the length of the flagellum (Langousis and Hill, 2014; Sunter and Gull, 2016; Taylor and Godfrey, 1969). The FAZ contains four specialized subpellicular microtubules (the microtubule quartet, MtQ, which initiates at the basal body and continues to the anterior cell tip), associated intracellular membranes contiguous with the endoplasmic reticulum (ER) and nuclear envelope (NE), and a filament system that connects to the axoneme and PFR via several membrane-spanning structures (Höög et al., 2012; Kohl and Bastin, 2005; Ralston et al., 2009; Sunter and Gull, 2016; Zhou et al., 2015). The flagellum beats in distinctive wave patterns that are adapted for movement in the bloodstream of mammalian hosts (Heddergott et al., 2012; Krüger et al., 2018; Rodríguez et al., 2009). Movement of BSF T. brucei is essential for cell viability and cytokinesis (Broadhead et al., 2006; Ralston and Hill, 2006), virulence (Shimogawa et al., 2018), development and pathogenesis (Langousis and Hill, 2014), and evasion of complement-mediated cell killing (Engstler et al., 2007; Heddergott et al., 2012).
The biogenesis of flagella requires the intraflagellar transport (IFT) machinery, which consists of an evolutionarily conserved suite of IFT proteins (Taschner and Lorentzen, 2016; van Dam et al., 2013), and kinesin and dynein motors (Prevo et al., 2017; Scholey, 2013). This machinery is under dynamic regulation to control flagellar structure and function (Ishikawa and Marshall, 2011). The IFT machinery moves cargoes along the axoneme in IFT trains (Cole et al., 1993, 1998; Kozminski et al., 1993; Lechtreck, 2015; Pigino et al., 2009), and mediates cargo entry into and exit from the flagellar compartment (Buisson et al., 2013; Dishinger et al., 2010; Kee et al., 2012; Verhey et al., 2011). Anterograde IFT transport is driven primarily by kinesin-2 family motor proteins (Prevo et al., 2017; Scholey, 2013), including two major subfamilies: heterotrimeric kinesin-2, comprising two different kinesin motor subunits (2A and 2B; KRP85 and KRP95, or KIF3A and KIF3B in vertebrates) (Cole et al., 1993; Kozminski et al., 1995) and the non-motor kinesin-associated protein (KAP, or KAP3A/3B in vertebrates) (Mueller et al., 2005; Wedaman et al., 1996); and in metazoans, homodimeric kinesin-2, which consists of two identical (2C; OSM3, or KIF17 in vertebrates) motor subunits (Prevo et al., 2017; Scholey, 2013; Signor et al., 1999). In metazoans, these two subfamilies have been shown to act both cooperatively and independently in IFT (Insinna and Besharse, 2008; Prevo et al., 2015; Snow et al., 2004). Kinesin-2 proteins also participate in extraciliary processes including transport of organelles, proteins, RNAs and virus cargoes, mitosis, cytokinesis, cell polarization, and cell adhesion and development (Prevo et al., 2017; Scholey, 2013).
In this study, we investigate the function of kinesin-2 proteins in BSF T. brucei. The T. brucei genome encodes ∼48 kinesins, including two kinesin-2 proteins (Berriman et al., 2005; Bertiaux et al., 2018a; Wickstead et al., 2010a), TbKin2a and TbKin2b. Our bioinformatic analysis suggests that TbKin2a and TbKin2b lack common sequence motifs that are broadly conserved in kinesin-2 proteins, and T. brucei lacks a gene encoding a homolog of the KAP subunit, which is required for heterotrimer function. We further found that both TbKin2a and TbKin2b contribute to flagellar biosynthesis in BSF T. brucei. Interestingly, silencing TbKin2a inhibits cell proliferation, cytokinesis and motility, whereas silencing TbKin2b does not. Moreover, TbKin2a localizes to the FAZ, suggesting it functions both within the flagellum and cell body. Thus, kinesin-2 proteins in trypanosomes play both canonical and unanticipated roles.
RESULTS
T. brucei has two divergent kinesin-2 proteins
The T. brucei brucei genome encodes the kinesin-2 proteins TbKin2a (Tb927.5.2090) and TbKin2b (Tb927.11.13920) (Fig. 1A), which were classified according to their motor domain (MD) sequences (Berriman et al., 2005; Wickstead and Gull, 2006; Wickstead et al., 2010a) using phylogenetic inference analysis (Goodson et al., 1994; Wickstead et al., 2010a) (see Fig. S1A for kinesin-2 phylogenetic tree from Wickstead et al., 2010a; see Table S1 for MD sequence identity comparisons). However, sequences in the kinesin-2 the neck-stalk-tail (NST) region, while not as well conserved as MD sequences (average of 17% NST sequence identity versus 54% MD sequence identity), are also important for kinesin function in metazoan species (De Marco et al., 2001; De Marco et al., 2003; Doodhi et al., 2009; Imanishi et al., 2006; Vukajlovic et al., 2011) (see Table S2 for identity comparison of NST sequences). Nevertheless, systematic analysis of kinesin-2 NST sequence motifs in a broad evolutionary context has been limited. We thus sought to further categorize kinesin-2 proteins based on NST sequences.
We compiled a phylogenetically diverse set of 81 kinesin-2 sequences from 25 species (Table S3A), which represent a broad cross-section of known phylogenetic diversity, and used the multiple Em for motif elicitation (MEME) tool suite to analyze these sequences (Fig. 1B–D; Figs S1 and S2, Table S3A,B). MEME identified a single amino acid motif consistent with a kinesin-2 neck domain at the beginning of the NST domain for all kinesin-2 proteins in our data set (Case et al., 2000; Vale and Fletterick, 1997). MEME also identified over 20 additional, statistically significant NST sequence motifs that can be sorted into distinct ordered motif groups and are broadly conserved across all kinesin-2-containing eukaryotic superfamilies from Metazoa to Excavata (Fig. 1D). Notably, motifs forming a primary (1°) motif group were common to almost all protist and metazoan taxa. A secondary (2°) motif group frequently associated with coiled-coil domains, as well as motifs that are specific to subgroups 2A and/or 2B, were also widely shared among protists and metazoa (Fig. 1B–D; Figs S1B,C; S2). The assignment of protist taxa encoding kinesin-2 into subgroups 2A and 2B was not observed in the phylogenetic analyses using MD sequences. MEME also assigned 2C motifs to metazoan kinesin-2C taxa, but such motifs were absent from protist taxa. The primary exception was a single 2C motif shared among kinetoplastids and metazoans, as discussed below (Fig. 1B–D; Figs S1 and S2).
The greatest divergence in NST sequence motifs occurs within the kinetoplastids, including T. brucei (Fig. 1B–D). Kinetoplastid kinesin-2 NST domains are longer, have more extensive predicted coiled-coil sequences, and lack most motif groups that are present in other protists and metazoans (Fig. 1; Fig. S2). Instead, the kinetoplastids have unique motif groups (knto/k) that are specific to these organisms. Parasititic kinetoplastid species (Julkowska and Bastin, 2009) and the free-living kinetoplastid Bodo saltans, lack the KAP subunit (k3) (Fig. 1B,D), which is required for kinesin-2 heterotrimer function and is present in all other organisms included in our analysis. Surprisingly, MEME assigned, to the kinetoplastid homologs of TbKin2b, a single, statistically significant kinesin-2C (OSM3) motif that is also present only in metazoan taxa (Fig. 1B–D,E; Figs S1 and S2). Together, these results suggest that kinetoplastid kinesin-2 proteins do not have a heterotrimeric structure consisting of a canonical 2A, 2B and KAP subunits, and raise the possibility that kinesin-2 proteins from kinetoplastid species may form homodimers and/or heterodimers. Thus, MEME analysis of NST sequences suggests that kinesin-2 proteins in T. brucei have diverged from such proteins in other organisms, raising questions about the extent to which their canonical functions are conserved in trypanosomes and whether novel functions have evolved.
TbKin2a distributes along the length of the flagellum, with stronger localization near the basal body
To assess the function of kinesin-2 proteins in T. brucei, we raised antibodies to a portion of the stalk domain. These specifically recognize the ∼130 kDa TbKin2a protein on immunoblots of whole-cell lysates (Fig. S3). We next assessed TbKin2a localization in BSF cells by immunofluorescence microscopy. TbKin2a localized in a punctate pattern along the lengths of both old and new flagella (Fig. 2A). During the cell cycle, TbKin2a staining intensity on flagella increased, and the relative staining between old and new flagella remained roughly equivalent, except that TbKin2a typically showed significantly enhanced staining on new flagella relative to old flagella during the one dividing kinetoplast, one nucleus (1K*1N) stage, when there is new flagellum growth. This reflects cytoskeletal remodeling at the posterior end associated with the emergence of the new flagellum (Gluenz et al., 2011; Ikeda and de Graffenried, 2012; Lacomble et al., 2010; Wheeler et al., 2013). We also determined the mean fluorescence intensity along full-length flagella. Using the peak intensity of DAPI-stained kinetoplast DNA as a posterior reference point (Fig. 2B), we determined that the flagellar intensity profile (on both old and new flagella) had two peaks at ∼0.7 µm and ∼2.1 µm, approximating locations near the basal body and flagellar pocket collar. Past the second peak, staining diminished to the anterior tip, where it increased slightly, with 65% of integrated staining intensity being confined to the proximal half of the flagellum.
TbKin2a also localized in the region of the kinetoplasts and basal bodies throughout their duplication cycle (Fig. 2C–F). There was extensive overlap between TbKin2a and the proximal end of basal bodies, as marked by the basal body marker BBA4 (Hoffmann et al., 2018; Schneider and Ochsenreiter, 2018; Woods et al., 1989) (Fig. 2G,H). TbKin2a and BBA4 colocalized with kinetoplastid DNA near the proximal ends of basal bodies (Fig. 2G,H). Thus, TbKin2a is positioned on the basal body adjacent to the kinetoplast.
TbKin2a colocalizes with IFT proteins primarily at the flagellum base
Although the canonical role of kinesin-2 proteins is in IFT (Lechtreck, 2015; Scholey, 2013), the divergent sequences of TbKin2a and TbKin2b raised the question of whether they retain their function in IFT in trypanosomes. In T. brucei, a full complement of canonical IFT complex B (anterograde; IFTB) and IFT complex A (retrograde; IFTA) and Bardet–Biedl syndrome (BBSome) proteins have been identified and shown to localize to the flagellum and basal body regions (Absalon et al., 2008; Adhiambo et al., 2009; Franklin and Ullu, 2010; van Dam et al., 2013). They have also been shown to interact and function in IFT (Bertiaux et al., 2018b; Bhogaraju et al., 2013; Blisnick et al., 2014; Buisson et al., 2013; Fort et al., 2016; Huet et al., 2014, 2019; Wachter et al., 2019). We co-stained for TbKin2a with IFTB marker IFT172 (Fig. 3A–D) or IFTA marker IFT144 (Fig. 3E). As previously reported in PCF cells (Absalon et al., 2008), in BSF cells these markers exhibited punctate staining along the flagellum and increased intensity near basal bodies. Along the flagellum, there was little colocalization between TbKin2a and either IFT172 or IFT144 (Fig. 3A,C,E,F; mean Mander's colocalization coefficients 0.164 and 0.278, respectively). However, we did observe clear patterns of colocalization of TbKin2a and IFT172 (Fig. 3A–D,F) and TbKin2A and IFT144 (Fig. 3E,F) in the region near the kinetoplasts and basal bodies, near the proximal ends of flagella (Fig. 3E,F; mean Mander's colocalization coefficients for TbKin2a–IFT172 and TbKin2a–IFT144 of 0.655 and 0.636, respectively). This suggests that TbKin2a might interact indirectly with IFT complex proteins primarily near the flagellum base.
TbKin2a localizes to the FAZ throughout the cell cycle
Given the tight association of the flagellum with the FAZ, we sought to distinguish whether TbKin2a also localizes to the FAZ. The flagellum can be detached from the cell-body side of the FAZ and cell body by detergent extraction, which permits separate visualization of the components localized to each structure (Robinson et al., 1991). Surprisingly, at all stages of the cell cycle, endogenous TbKin2a colocalized with the intracellular FAZ marker L3B2 (Kohl et al., 1999) in a continuous punctate pattern from the initiation of the FAZ (start of L3B2 staining) near the flagellar pocket collar (Lacomble et al., 2009, 2010) to the cell anterior tip (Fig. 4A). (We did not observe IFT172 colocalization with the FAZ, data not shown.)
Because detergent extraction significantly diminished TbKin2a staining on flagella (as also observed with IFT proteins; Absalon et al., 2008), we sought to identify conditions to separate the flagellum and FAZ following fixation and without detergent extraction. We observed that the flagellum and FAZ were sometimes partially separated due to shear forces during mounting. Under these circumstances, TbKin2a could always be seen on both structures (Fig. 4B,C). We could also use deliberate controlled shearing to induce similar separation of the flagellum and FAZ post-fixation, and under these circumstances, TbKin2a could similarly be seen on both structures (Fig. 4D). To further confirm these results, we triggered flagellar detachment in living cells through Ca2+ chelation (Vickerman, 1969), either by treatment with citrate or EGTA. This treatment resulted in cells with detached flagella, including 1K1N cells with a full-length FAZ and one flagellum that is mostly detached from the cell body, except at the flagellar base. In these cells, TbKin2a was localized to both the flagellum and FAZ (Fig. 4E; Fig. S4).
Finally, to rule out the possibility that the anti-TbKin2a antibody non-specifically recognizes FAZ components, we engineered BSF cells to express C-terminally Myc-tagged TbKin2a (TbKin2a–Myc) (Fig. S5A) and stained these cells with anti-Myc antibody. TbKin2a–Myc also colocalized with the FAZ marker L3B2 (Fig. S5B). TbKin2a staining continued to the basal body area, separate from its staining on flagella, proximal to the FAZ initiation point. Collectively, these results indicate that TbKin2a associates with the FAZ as well as with the flagellum, most likely on the MtQ, which is the sole intracellular cytoskeletal element know to connect from the basal body to the FAZ.
Silencing of TbKin2a inhibits proliferation of BSF T. brucei, whereas silencing of TbKin2b does not
To determine the importance of kinesin-2 proteins in BSF T. brucei, we used RNA interference (RNAi) to silence expression of TbKin2a, TbKin2b or both. TbKin2a mRNA was diminished at 24 h and 48 h post induction (hpi) of dsRNA expression in TbKin2a-silenced cells, but remained unaffected in TbKin2b-silenced cells, as assessed by RT-PCR (Fig. 5A) and quantitative (q)RT-PCR (Fig. 5B). Similarly, TbKin2b mRNA was diminished at 24 hpi and 48 hpi in TbKin2b-silenced cells, but was unaffected in TbKin2a-silenced cells (Fig. 5A,B). Both mRNAs were diminished in TbKin2a2b-silenced cells. For TbKin2a, protein expression remained at 24 hpi but was strongly diminished after 48 hpi (Fig. 5A; Fig. S3B).
Interestingly, TbKin2a-silenced cells proliferated 50–70% more slowly than uninduced cells after 24 hpi and ceased proliferation by 48 hpi (Fig. 5C). For TbKin2b-silenced cells, proliferation did not decrease significantly by 72–96 hpi (Fig. 5C). Silencing both TbKin2a and TbKin2b together resulted in a cessation of proliferation with kinetics similar to single TbKin2a knockdown (Fig. 5C). Thus, cell proliferation in BSF T. brucei requires TbKin2a, but proliferation is not affected by RNAi silencing of TbKin2b.
Evidence of the essential role of TbKin2a in cell proliferation led us to assess its role in cell cycle progression. We first counted the numbers and observed the morphology of nuclei and kinetoplasts at various time points after RNAi induction (Fig. 5D). Beginning at 24 hpi, the proportion of normal 1K1N (1 kinetoplast, 1 nucleus), 2K1N, and 2K2N cells decreased progressively, while abnormal cells with multiple nuclei and kinetoplasts (xK†yN†) or with one abnormally large nucleus and multiple kinetoplasts (xK†1N†), increased progressively (Fig. 5D). We next observed the effect of TbKin2a silencing on the cell cycle state by FACS (Fig. 5E). From 24 hpi to 48 hpi, the percentage of cells with 2C DNA content decreased, whereas the percentage of cells with 4C DNA content increased. After 48 hpi, there was a significant increase in the proportion of cells with ≥4C DNA content. This suggests that nuclear DNA synthesis continued without complete nuclear separation and/or cytokinesis.
At the morphological level, cells induced for 24 h or longer became much larger than uninduced cells, accumulating multiple nuclei, flagella (Fig. 5F,G) and FAZ (Fig. 5H). Typically, flagella and FAZ clustered at the anterior ends of undivided cells, and there was little evidence of cleavage furrow formation or progression (Fig. 5F,G,H). These results suggest that TbKin2a is important for cytokinesis, which initiates at the anterior end (Sherwin and Gull, 1989).
In addition, at 24 hpi, 48 hpi and 72 hpi, TbKin2a and TbKin2a2b-silenced cells showed nuclei that were often poorly defined or separated, resulting in large undivided or partially divided nuclear masses with numerous nucleoli (Fig. 5G–I). Although kinetoplast DNA replication appeared to continue, ∼8–10% of kinetoplasts appeared to be partially separated or not separated (Fig. 5G–I). Basal body duplication and separation were also ongoing, but we observed multiple TbKin2a- and TbKin2a2b-silenced cells with basal bodies having no associated kinetoplast DNA (Fig. 5J). These results suggest that TbKin2a contributes to separation of nuclei and kinetoplasts, and might play a role in the association of basal bodies with kinetoplasts.
TbKin2a and TbKin2b act additively in forming full-length flagella, but only TbKin2a is required for motility
Considering the established role of kinesin-2 in flagellar assembly in other organisms, we evaluated the effect of silencing TbKin2a and/or TbKin2b on flagella length. Surprisingly, given their sequence divergence from other kinesin-2 proteins, silencing either TbKin2a or TbKin2b for 72 h caused a decrease in flagellum length of 22% and 21%, respectively (Fig. 6A). Silencing both together caused an approximately additive 42% decrease in flagellar length. We could not observe whether a subpopulation of flagella remained long, in contrast with recent observations for silencing IFT proteins in PCF T. brucei (Bertiaux et al., 2018a; Fort et al., 2016), as the cytokinesis defect and buildup of polyploid cells caused by TbKin2a or TbKin2a2b silencing precludes separate identification of old and new flagella. This indicates that TbKin2a and TbKin2b act additively to build and/or maintain the flagellum in BSF T. brucei.
To determine whether the defects in flagellar length and IFT protein localization caused by TbKin2a silencing were linked with gross ultrastructural defects in the flagellum, we examined TbKin2a RNAi cells at 48–72 hpi using transmission electron microscopy (TEM) (Fig. 6B–D). Although cell morphology was dramatically perturbed, we observed infrequent gross structural anomalies in flagella and no FAZ defects (Fig. 6B,C). In wild-type and uninduced cells, we observed localization of IFT cargoes primarily on axoneme doublets 3–4 and 7–8, consistent with what was seen in PCF cells (Bertiaux et al., 2018a). In TbKin2a-silenced cells, 13% (n=55 images) had abnormal accumulations of material between the flagellar membrane and axoneme outer pairs 3–4 or 8–9 [the location of IFT cargo trains in T. brucei (Absalon et al., 2008)], whereas <3% (n=93 images) of control cells had such accumulations. Moreover, the nuclei of TbKin2a-silenced cells frequently displayed irregular shapes, areas of NE disorganization, and electron dense plaques of various sizes on the inner periphery of the NE (Fig. 6C,D). The plaques were similar in number and location to chromatin plaques observed at the periphery of nuclei in T. brucei rhodesiense, beginning in late mitosis and persisting into interphase (Farr and Gull, 2012; Vickerman and Preston, 1970). This ultrastructural evidence is consistent with the conclusion that TbKin2a plays a role in IFT and suggests that it may also play a role in late cell-cycle timing as well as NE and chromatin organization.
To measure the impact of TbKin2a and TbKin2b silencing on motility, we adapted a previously established sedimentation assay for PCF T. brucei in which the rate of sedimentation is inversely proportional to motility capacity (Bastin et al., 1999; Ralston and Hill, 2006). Sedimentation behavior was assessed over 4–8 h, beginning at 18 hpi, 24 hpi and 36 hpi. At 18–22 hpi, sedimentation rates for induced and uninduced cells were statistically equivalent (data not shown). However, after 24 hpi, cells silenced for TbKin2a or TbKin2a2b sedimented significantly faster than uninduced controls at every time point (Fig. 6E). In contrast, silencing TbKin2b alone did not result in faster sedimentation relative to controls. We conclude that TbKin2a silencing, but not TbKin2b silencing, results in a significant, progressive decrease in motility.
Finally, we evaluated the kinetics of onset of the TbKin2a RNAi phenotype by identifying abnormal cell morphologies as they formed over the first 26 hpi. We observed small numbers of abnormal cells beginning 16 hpi, and over the next 4–8 hpi abnormal cells accumulated such that, by 26 hpi, over 20% of the cell population had abnormal morphologies. The most frequent abnormal early phenotype observed was clustered or bundled flagella at the anterior cell end within abnormally large cells (Fig. 6F). Thus, the earliest phenotype associated with silencing TbKin2a is an apparent failure to initiate cytokinesis.
DISCUSSION
Kinesin-2 proteins perform an important and conserved function in IFT and flagellar biogenesis of primary and motile cilia (Lechtreck, 2015; Prevo et al., 2017; Scholey, 2013). Kinesin-2 proteins have also been implicated in bidirectional ER to Golgi transport (Brown et al., 2014; Stauber et al., 2006), endosomal trafficking (Granger et al., 2014), transport in neurons (Hirokawa et al., 2010), cilium-based signaling (Goetz and Anderson, 2010), chromosome segregation (Haraguchi et al., 2006; Miller et al., 2005) and cytokinesis completion (Brown et al., 1999; Fan and Beck, 2004). Here, we report that the kinesin-2 proteins TbKin2a and TbKin2b in BSF T. brucei function in flagellar biogenesis, but only TbKin2a appears to be crucial for cell proliferation. In addition, TbKin2a is found on the FAZ, suggesting a possible role in FAZ-based transport. Thus, kinesin-2 proteins in trypanosomes are likely to perform both canonical and trypanosome-specific roles.
Trypanosome kinesin-2 proteins have divergent NST sequences
The results of our MEME analysis complements previous kinesin bioinformatics studies (Berriman et al., 2005; Wickstead et al., 2010a) by identifying conserved sequence motifs in the non-motor-domain NST portions of kinesin-2 proteins. Unexpectedly, we found that kinesin-2 proteins contain over 20 conserved NST motifs, with several key motif groups, and are broadly shared across all superfamilies of flagellated eukaryotes. Several of these motifs had been previously recognized within a narrower phylogenetic context (De Marco et al., 2001; De Marco et al., 2003; Doodhi et al., 2009; Imanishi et al., 2006; Vukajlovic et al., 2011), while others, in particular NST motifs specific to subgroups 2A, 2B and 2C, appear to be new. Some motifs and motif groups were common to almost all taxa encoding kinesin-2, while others were specific for the taxa encoding heterotrimeric kinesins 2A and 2B or the homodimeric kinesin 2C.
Although the IFT machinery is highly conserved within the kinetoplastids (van Dam et al., 2013), it was surprising that the kinetoplastid taxa did not share common kinesin-2 NST motif groups and had little evidence of conserved individual motifs (Fig. S2). Kinetoplastids are also the only kinesin-2-containing organisms identified that do not encode non-motor KAP homologs in their genomes (Julkowska and Bastin, 2009). The free-living kinetoplastid B. saltans also does not encode a KAP homolog, suggesting that it was not lost as a result of parasitism. Our MEME analysis and KAP genomic data do not support bioinformatic assignment of kinetoplastid kinesin-2 proteins as orthologs of the canonical heterotrimeric kinesin-2A or 2B forms. However, a kinetoplastid kinesin-2 protein did share a single significant kinesin-2C motif with those found in Metazoa, which was the sole example of a homodimeric kinesin-2C motif among non-holozoan single-celled organisms. Together with the markedly different phenotypes caused by TbKin2a versus TbKin2b RNAi silencing in BSF T. brucei, these data suggest that kinesin-2 proteins in T. brucei do not form canonical heterotrimers. Instead, the kinetoplastid kinesin-2 proteins represent a distinct subgroup that differ significantly in sequences and motifs, may form homodimers or heterodimers, and may differ in function compared with kinesin-2 proteins in other organisms.
Trypanosome kinesin-2 proteins are important for flagellum biosynthesis
We found that silencing TbKin2a or TbKin2b results in a ∼20% decrease in flagellum length, while silencing both causes a decrease of 40% (although flagella are present). Moreover, TbKin2a localizes along the flagellum, with a concentration on basal bodies and the proximal flagellum, similar to the localization of the Chlamydomonas reinhardtii kinesin-2 FLA10 (Cole et al., 1998; Deane et al., 2001; Vashishtha et al., 1996). TbKin2a also colocalizes with both the IFTB protein IFT172 and IFTA protein IFT144 near basal bodies. Silencing TbKin2a decreased staining of both IFT172 and IFT144 on the flagellum and decreased the localization of IFT144, but not IFT172, to the basal body. Moreover, flagella in TbKin2a-silenced cells showed accumulation of electron-dense material along axonemes, but otherwise did not show gross abnormalities. [We could not always resolve fine structural elements such as inner dynein arms, which require kinesin-2 for transport in C. reinhardtii (Piperno and Mead, 1997; Piperno et al., 1996).] These data suggest that TbKin2a and TbKin2B participate in IFT.
One distinction between TbKin2a and TbKin2b that is of potential significance is the presence of a putative C-terminal nuclear localization signal (NLS) sequence in TbKin2a, but not in TbKin2b [NLS predicted using NucPred (Brameier et al., 2007) and MultiLoc2 (Blum et al., 2009)]; this sequence might also function as a ciliary localization signal sequence (CLS) (Verhey et al., 2011). For the homodimeric kinesin-2 protein KIF17, an NLS/CLS mediates entry into the ciliary compartment (Dishinger et al., 2010; Kee et al., 2012). Thus, TbKin2a and TbKin2b may play different roles in cargo entry into flagella. Investigations in C. elegans indicate that heterotrimeric kinesin-2A/2B is primarily responsible for cargo entry into the flagellum transition zone. Following cargo entry, kinesin-2A/2B gradually undocks while homodimeric kinesin-2C docks in the proximal segment. Finally, kinesin-2C becomes the primary transport motor in the distal segment (Prevo et al., 2015). By analogy, TbKin2a and TbKin2b may play different roles in cargo transport at different flagellar regions.
In many other organisms, kinesin-2 proteins are essential for flagellum biogenesis and maintenance (Lechtreck, 2015; Prevo et al., 2017; Scholey, 2013). However, in BSF T. brucei, silencing kinesin-2 proteins resulted in partial but not complete inhibition of flagellum biosynthesis and maintenance. This may be due to incomplete RNAi silencing, slow protein turnover, the presence of other kinesins with redundant roles or increasingly pleiotropic effects of RNAi over time or other processes in BSF cells such as cell cycle progression, which result in cell death before flagellar shortening is complete (Blaineau et al., 2007; Broadhead et al., 2006; Chan and Ersfeld, 2010; Demonchy et al., 2009; Marande and Kohl, 2011; Wickstead et al., 2010b). Interestingly, in PCF T. brucei IFT is active in both new and old flagella, but silencing either anterograde IFTB or retrograde IFTA components causes a failure in new flagellum biosynthesis (although with somewhat different phenotypes) but no defect in old flagellum maintenance (Fort et al., 2016). Another recent study demonstrated that individually silencing TbKin2a or TbKin2b had no effect on flagellum length in PCF, but silencing both caused shorter flagella (Bertiaux et al., 2018a), suggesting a redundant function in IFT. Our results are consistent with both individual and redundant roles for kinesin-2 proteins in anterograde IFT during new flagellum biosynthesis in BSF T. brucei.
TbKin2a is essential for cell proliferation
Interestingly, silencing TbKin2a expression caused a cessation of cell proliferation beginning at ∼24 hpi, whereas silencing TbKin2b expression had no apparent effect on cell proliferation. Silencing TbKin2a also caused a failure in cytokinesis, with the accumulation of enlarged cells with multiple nuclei, kinetoplasts, basal bodies, FAZ and flagella, as well as decreased cell motility. It is surprising that only TbKin2a is important for proliferation, cytokinesis and motility, although both TbKin2a and TbKin2b contribute to flagellar length. This suggests that cells can withstand flagellar length decreases of at least 20%, and that shorter flagella in BSF cells can be functional, perhaps through a compensatory mechanism such as decreased cell size, as in PCF T. brucei (Kohl et al., 2003).
The cell proliferation defects caused by silencing TbKin2a in BSF cells are similar to the defects caused by RNAi silencing of other flagellar proteins, FAZ proteins and basal body proteins (Broadhead et al., 2006; LaCount et al., 2002; Morris et al., 2001; Ralston et al., 2006), as well as cell-cycle-related proteins, signaling proteins and cell surface proteins (Farr and Gull, 2012; Hammarton et al., 2007b). The commonality of phenotypes is specific to BSF cells and may be due to the fact that silencing these factors directly or indirectly impacts cytokinesis (Hammarton et al., 2007b; Hughes et al., 2013; Oberholzer et al., 2011; Wheeler et al., 2013; Zhou et al., 2014).
It has been hypothesized that cell proliferation and cytokinesis defects in BSF trypanosomes can result from inhibited flagellar beating (Broadhead et al., 2006; Ralston et al., 2006). However, defects in flagellar beating may not explain the cell proliferation and cytokinesis phenotypes observed in TbKin2a-silenced cells for several reasons. First, normal flagellar beating is not essential for cytokinesis, because the cell proliferation and cytokinesis defects caused by silencing dynein light chain 1 by RNAi can be rescued by the expression of site-specific mutants that are still defective in flagellar motility (Kisalu et al., 2014; Ralston et al., 2011). Second, silencing TbKin2a did not cause gross ultrastructural abnormalities in flagella or the flagellar pocket other than abnormal IFT train accumulation, in contrast to the major defects caused by silencing other flagellar proteins (Broadhead et al., 2006; Farr and Gull, 2009). Third, we find that following TbKin2a RNAi induction, the kinetics of onset of the cytokinesis defect precedes the defect in cell motility. For example, at 16–20 hpi, abnormal cells with bundled flagella and FAZ at their anterior ends are observed, suggesting a defect in cytokinesis initiation. Abnormal cells are also observed connected at their posterior ends, suggesting a defect in cell scission. At 24 hpi, cell proliferation is affected, and there is a decrease in motility thereafter, although the flagella in the abnormal cells continue to beat. These considerations suggest that TbKin2a may play a role in cytokinesis initiation that is independent of any role in flagellar motility. This may be due to its involvement in transporting cargoes into flagella that themselves are important for cytokinesis or its presence on the FAZ, as discussed below.
Implications of TbKin2a localization on the FAZ
It is intriguing that TbKin2a also localizes to the FAZ. However, TbKin2a does not appear to be important in the construction, maintenance or structure of the FAZ because we did not observe abnormalities in the FAZ or flagellar detachment in TbKin2a- or TbKin2a2b-silenced cells. Instead, we propose that TbKin2a interacts with the FAZ via the MtQ and may transport other cargoes or facilitate intracellular signaling, including those essential for cell division. This is the simplest model, given the ability of TbKin2a to bind to microtubules, and that TbKin2a staining is present between the basal body and FAZ initiation site in the same location as the MtQ (Lacomble et al., 2009, 2010). Because MtQ microtubules are uniformly oriented with their plus ends towards the cell anterior (Gull, 1999), antiparallel to the remainder of the subpellicular array (Robinson et al., 1995), they represent a potential track for moving cargoes along the anterior–posterior cell axis.
We speculate that cargoes transported by TbKin2a along the FAZ could include both organelles and cell cycle regulatory molecules. The FAZ is linked with other single-copy structures including the kinetoplast, tripartite attachment complex (TAC), basal body, flagellum, flagellar pocket collar, and bilobe structure with ER and Golgi exit sites (Gadelha et al., 2009; Gheiratmand et al., 2013; Gluenz et al., 2011; Lacomble et al., 2009, 2010; Sunter et al., 2015; Zhou et al., 2015). TbKin2a may be involved in transporting components involved in these interactions. For example, basal bodies and kinetoplasts are connected by the TAC (Ogbadoyi et al., 2003), and basal body duplication and separation are necessary for kinetoplast segregation (Robinson and Gull, 1991). We observed defects in kinetoplast separation and basal body to kinetoplast attachment in TbKin2a-silenced cells, suggesting that TbKin2a may participate in TAC assembly or maintenance. TbKin2a may also be important for the transport of structural or regulatory proteins that are important for cell cycle regulation as well as duplication and segregation of many of the above-mentioned components, which include the centrins 1, 2 and 4 (Selvapandiyan et al., 2007; Shi et al., 2008; Wang et al., 2012), as well as polo-like kinase (TbPLK) (de Graffenried et al., 2008, 2013; Ikeda and de Graffenried, 2012; Li et al., 2010; Umeyama and Wang, 2008). Notably, TbPLK is transported along the MtQ to the anterior tip of the cell (Ikeda and de Graffenried, 2012; Li et al., 2010; Sun and Wang, 2011; Yu et al., 2012), where it is required for cytokinesis initiation (Hammarton et al., 2007a; Kumar and Wang, 2006). Recent studies have also identified new proteins that are associated with the FAZ and control various aspects of T. brucei cytokinesis (Hilton et al., 2018; Sinclair-Davis et al., 2017; Zhang et al., 2019; Zhou et al., 2018). Future experiments will be aimed at determining whether TbKin2a and TbKin2b transport distinct cargoes along the flagellum and/or FAZ, and how cargo transport contributes to flagellum construction, organelle transport and positioning, and cytokinesis.
MATERIALS AND METHODS
MEME suite analysis
Multiple EM for Motif Elicitation (MEME) web-based software (Bailey and Elkan, 1994; Bailey and Gribskov, 1998; Bailey et al., 2006, 2009) (versions 3.91-4.11) was used to analyze the NST domains of putative kinesin-2 proteins. Using taxa encoding kinesin-2 identified in a phylogenetic analysis of kinesin motor domain sequences from 45 diverse organisms (Wickstead et al., 2010a) as the starting point, we carried out more than 80 independent MEME computational runs including multiple combinations of kinesin-2 NST data subsets and input parameters with multiple controls. See Table S3 for a reconciliation of kinesin-2 sequences used in our MEME NST analyses with and kinesin-2 sequences used in a prior motor domain analysis (Wickstead et al., 2010a). To ensure that all NST domains consistently included a full neck region, which has been shown to evolve independently of the motor domain (Case et al., 2000; Vale and Fletterick, 1997), our NST domain for each taxon began with the final three amino acids (positions 343–345) of the PFAM-defined motor domain sequence specification (PFAM00225). MEME results were sensitive to changes in inputs and assumptions including composition, length and number of protein sequences, expected motif distribution and run parameters. To overcome this sensitivity, many runs of kinesin-2 sequences, as well as control sequences, were conducted until consistent result patterns were observed. Our controls included running a full MEME analysis of: kinesin-1 family NST sequences; kinesin-1, -3 and -4 family neck domain sequences; kinetoplastid and non-kinetoplastid kinesin NST sequences; and scrambled kinesin-2 NST sequences (NST sequences are found in Table S4). Overall results were generally robust and consistent but as already noted, could vary from run to run. Kinesin-2 NST motif results presented here are from a single run (MEME 83) that we judged to be highly consistent with overall result patterns from about eight final MEME runs. All MEME statistics presented were generated by standard MEME programs (Bailey and Elkan, 1994; see Figs S1, S2, Table S3).
Cell culture and RNA interference
BSF T. brucei brucei Lister 427-derived cell line 90-13 (Wirtz et al., 1999) was cultured and maintained in HMI-9 media. Media was initially supplemented with 10% Serum Plus (MilliporeSigma, Burlington, MA) and 10% tetracycline-tested, heat-treated fetal bovine serum (FBS) (R&D Systems, Minneapolis, MN), and later supplemented with 15–20% FBS only (complete medium), as otherwise described previously (Li and Wang, 2006; Tu and Wang, 2004), with splitting of the cultures at final densities of ∼1.0×106 cells/ml. For RNAi, DNA corresponding to the 380-bp T. brucei gene Tb927.5.2090 (TbKin2a) or the 462-bp gene Tb927.11.13920 (TbKin2b) (selected using RNAit software; Redmond et al., 2003) were amplified by PCR from genomic DNA (primers listed in Table S5). For RNAi of TbKin2a or TbKin2b, PCR fragments were ligated into the XhoI and HindIII sites of pZJM (Wang et al., 2000). For RNAi of both TbKin2a and TbKin2b, the TbKin2b PCR product was ligated into the XbaI and XhoI sites in pZJM that already contained TbKin2a. All DNA sequences were verified. Transfections (Burkard et al., 2007; Li and Wang, 2006) and selection of monoclonal transformants were performed using agarose plates as described previously (Carruthers and Cross, 1992; Tan et al., 2002). For induction of RNAi, monoclonal transformants were cultured with 1 µg/ml tetracycline for 0-96 h as indicated. For growth curves, live cells were counted using a hemocytometer.
Antibody generation
A portion of the TbKin2a gene encoding amino acids 391–696 plus a C-terminal 6×His tag (TbKin2a.391-696-His) was amplified by PCR using primers listed in Table S5, ligated into the NcoI and NotI sites in pET22b(+) (Novagen, Madison, WI), and verified by DNA sequencing. Recombinant TbKin2a.391-696-His was expressed in E. coli BL21-CodonPlus-RP cells (Agilent Technologies, Cedar Creek, TX) after induction with 400 µM IPTG at 37°C for 2 h. The protein was purified by Ni-NTA (Qiagen, Valencia, CA) affinity chromatography followed by gel filtration chromatography on a Superdex 75 column (GE Healthcare, Piscataway, NJ). To generate anti-TbKin2a antibody, rabbits were immunized by Covance Inc. (Princeton, NJ) using purified TbKin2a.391-696-His. Antibody was affinity purified using standard procedures.
Engineering T. brucei to express TbKin2a–Myc
To generate a T. brucei strain expressing Myc-tagged TbKin2a under the control of its endogenous promoter, we used the long primer PCR transfection method described for C-terminal fusion protein expression in bloodstream form T. brucei (Dean et al., 2015), except that plasmid pPOTv7 was used as template DNA for PCR amplification. pPOTv7 encodes a 10× Myc tag at the C-terminus and the blasticidin resistance bsr gene, rather than the hygromycin resistance hpt gene that is in pPOTv4. For forward long primer, we used final 80 nucleotides prior to the stop codon of the Tb927.5.2090 gene, followed in frame by the applicable 18 nucleotide pPOTv4/v7 annealing sequence (5′-GGTTCTGGTAGTGGTTCC-3′). For the reverse primer, we used the reverse complement of the first 80 nucleotides (without stop codon) of the 3′ untranslated region of Tb927.5.2090 gene followed in frame by the reverse complement form of the applicable 20 nucleotide pPOTv4/v7 annealing sequence (5′-CCAATTTGAGAGACCTGTGC-3′). Primers were synthesized by Integrated DNA Technologies (Coralville, IA). PCR was performed using Phusion HF DNA Polymerase (New England BioLabs, Ipswich, MA). The DNA sequence of the PCR products was confirmed. PCR products were ethanol precipitated and resuspended in water. For electroporation of T. brucei BSF 90-13 we used the AMAXA Nucleofector (Lonza, Basel, Switzerland) with program X-001 for BSF T. brucei, as specified in Burkard et al. (2007), except that we used the transfection buffer specified in Schumann Burkard et al. (2011). T. brucei cells were cultured at 37°C in HMI-9 medium containing 10% (v/v) heat-inactivated fetal bovine serum, with 2.5 μg/ml neomycin and 5 μg/ml hygromycin. After transfection, the cells were selected by serial dilution with the addition of blasticidin at 10 μg/ml.
Western blotting
For western blotting of cell lysates from wild-type BSF 90-13 cells and TbKin2a RNAi (induced and uninduced) cells, whole-cell lysates were boiled in 1× SDS loading buffer (200 mM Tris-HCl pH 6.8, 8% SDS, 0.3% Bromophenol Blue, 30% glycerol and 15% β-mercaptoethanol), separated by SDS-PAGE, and transferred to a nitrocellulose membrane. After a 1 h blocking step in TBS-T (20 mM Tris pH 8.0, 150 mM NaCl, 0.02% Tween 20) with 5% dry milk, membranes were incubated with primary antibodies in TBS-T plus 5% dry milk as follows: affinity-purified anti-TbKin2a antibody at a dilution of 1:10,000; YL 1/2 rat monoclonal anti-tyrosylated α-tubulin antibody (cat. no. MAB1864-I, MilliporeSigma, Burlington, MA) at a dilution of 1:100; KMX mouse anti-β-tubulin antibody (Birkett et al., 1985) at a dilution of 1:100. After three washes in TBS-T+5% dry milk, the membrane was incubated with applicable anti-rabbit, anti-rat or anti-mouse-IgG secondary antibodies conjugated to horseradish peroxidase (Thermo Fisher Scientific) at the manufacturer recommended dilutions and washed twice for 10 min in TBS-T plus 5% dry milk and twice for 5 min in PBS (137 mM NaCl, 2.7 mM KCl, 10 mM Na2HPO4, 2 mM KH2PO4, pH 7.4), and then blots were imaged.
For BSF trypanosomes expressing TbKin2a–Myc, whole-cell protein lysates were prepared and subjected to SDS-PAGE as described above, and proteins were transferred to nitrocellulose membranes. After a 1 h blocking step in TBS-T plus 5% dry milk, membranes were incubated with primary antibodies in TBS-T plus 5% milk as follows: mouse anti-Myc 9E10 (Evan et al., 1985) at a dilution of 1:1000; rabbit anti-TbKin2a at a dilution of 1:10,000. After three washes in TBS-T plus 5% dry milk, the membranes were incubated with secondary antibodies as follows: Alexa Fluor 680 anti-mouse-IgG (ThermoFisher Scientific, Waltham, MA) and AlexaFluor 790 anti-rabbit-IgG (ThermoFisher Scientific), both at 1:10,000 dilution. Membranes were washed twice in TBS-T plus 5% dry milk and twice for 5 min in PBS. Blots were imaged using the Odyssey infrared imaging system (Li-Cor Biosciences, Lincoln, NE).
RT-PCR and qRT-PCR
For all RT-PCR and qRT-PCR studies, total RNA (from at least two independent biological replicates) was extracted and isolated individually for each experiment from freshly spun-down, unwashed T. brucei BSF RNAi cells (induced or uninduced) using either TRI Reagent or RNAzol [Molecular Research Center (MRC), Cincinnati, OH]. RNA for semi-quantitative RT-PCR was purified after isolation using MRC RNA precipitation protocols, including a DNase treatment step using TurboDNAse (Thermo Fisher Scientific). For qRT-PCR from RNAi cells, RNA extracted and isolated as above from n=2 (TbKin-2a), n=2 (TbKin2b), and n=2 (TbKin2a2b) independent biological replicates was individually purified either as above, or by immobilizing extracted and isolated RNA on silica columns [Direct-zol, Zymo Research (ZR), Carlsberg, CA], treating with DNase, and purifying, all as described in the manufacturer's protocols (except that an additional wash step of 100% ethanol was included) and as a final step, eluted with pure DEPC water. Purified RNA in DEPC water was analyzed using a Nanodrop spectrophotometer (ThermoFisher Scientific) and stored at −80°C. For qRT-PCR, purified RNA integrity for each sample was confirmed (Schroeder et al., 2006) using an Agilent Bioanalyzer 2100 (Agilent Technologies, Santa Clara, CA) by the UC Berkeley QB3 Functional Genomics Laboratory. All primers were from Integrated DNA Technologies (Coralville, IA). For semi-quantitative RT-PCR, first-strand cDNAs were generated from RNA using SuperScript III First-Strand Synthesis System (ThermoFisher Scientific) with oligo(dT) primers, and PCR was performed using first-strand cDNA and with gene-specific primers. qRT-PCR used a two-step procedure with first-strand cDNAs generated from purified RNA using a Superscript VILO cDNA Synthesis Kit (ThermoFisher Scientific). We evaluated T. brucei potential reference standard genes as recommended previously (Brenndörfer and Boshart, 2010). Only one gene, TERT (telomerase reverse transcriptase, Tb927.11.10190), had minimal expression variation under our experimental conditions and was selected as sole reference standard. All qRT-PCR work was performed in 96-well plates using either Applied Biosystems (ABI) 7500 Fast Real-Time PCR System and iTaq SYBR Green Supermix (ThermoFisher Scientific); or Roche LightCycler96 System using iQ SYBR Green Master Mix (Bio-Rad, Inc., Richmond, CA) using separate wells for reference and target reactions. Gene-specific primers for qRT-PCR were designed using ABI Primer Express Version 2.0 software and/or Primer3 software (Untergasser et al., 2012) and sequences are in Table S5. Amplification efficiency of each gene/primer combination was confirmed over a 4- to 5-log dilution series. Relative gene expression for qRT-PCR was determined for each biological experiment, and results were aggregated using the 2−ΔΔCT method (Schmittgen and Livak, 2008).
Fluorescence-activated cell sorting
Cell samples for fluorescence-activated cell sorting (FACS) were prepared as described previously (Tu and Wang, 2004). FACS was carried out using an Epics XL flow cytometer (Beckman Coulter, Brea, CA). Data were analyzed using FlowJo 4-5 software (FlowJo LLC, Ashland, OR; a subsidiary of Becton, Dickinson and Company).
Motility assays
At experiment initiation, cells were centrifuged and transferred to 1–10 ml of fresh settling medium [a 1:1 (v/v) mix of heat-treated tetracycline-free fetal bovine serum (R&D Systems, Minneapolis, MN) and regular complete HMI-9 medium (Engstler et al., 2007)] at starting densities of ∼0.5×105–3×105 cells/ml, with or without tetracycline at 37°C with 5% CO2. Using previously described methods, including centrifuging live cells and replacing medium at 12–24 h intervals (Hesse et al., 1995), uninduced and induced cells were grown in settling medium for up to 48 h and reached final densities of up to∼5×106 cells/ml. At 18, 24, 36 and 48 h after growth in the settling medium and RNAi induction, cells were counted and 1 ml of cells was transferred into each of three grouped cuvettes per cell line (uninduced and induced) and sealed with sterile, liquid impermeable and gas-permeable film. Cuvettes were agitated and the initial A600nm at 0 min was measured using a SpectraMax M2 multi-detection reader (Molecular Devices, Sunnyvale, CA) or Genesys 10 spectrophotometer (Thermo Fisher Scientific, Waltham, MA). Cuvettes were incubated at 37°C with 5% CO2, and at each subsequent time point (every 1–2 h up to 8 h), cuvettes were removed from the incubator. A600nm measurements were obtained and changes in absorbance were calculated.
Methods for separating flagella from FAZ
For controlled shearing of flagella from cell bodies, we devised a simple parallel-plate flow chamber device and shearing procedure to apply measured, consistent shear forces. 24×40 mm poly-L-lysine-coated #1.5H high-precision coverslips (Paul Marienfeld GmbH & Co. KG, Lauda-Königshofen, Germany) were pre-weighed dry, then fully fixed cells (see staining for fluorescence microscopy) were added, coverslips were mounted onto a cleaned glass slide with 50–120 µl ProLong Gold antifade (Invitrogen), and the assembly was placed into a mounting slot in the device base, which had been levelled and clamped in place and maintained at room temperature. Coverslips were re-weighed after each step above. A 25×50 mm glass slide top-plate, with or without attached weights (total top plate mass as used was 7.55–10.52 g), was loaded carefully onto the coverslip, flush along the back side, between device guides. All masses were determined using a single, calibrated, fully-enclosed Mettler Toledo AB54-S analytical balance. Using a stopwatch and pusher rod placed directly at the back of the coverslip, the coverslip and top plate assembly was pushed by hand in a single direction at an approximately constant velocity of ∼0.7–1 mm/s for 4–5 s over a measured distance of ∼4–5 mm. Immediately before and after the above-mentioned steps, clean, dry pre-weighed Kimwipes (Kimberly-Clark, Roswell, GA) held in micro-forceps were used to blot-up excess liquid and immediately reweighed, with all before and after wipe masses recorded and weights adjusted. Shear-treated coverslips were then allowed to set in dark at 4°C for 24 h, then sealed as described in the section below.
It was previously noted that BSF T. brucei cells grown in citrated blood to chelate Ca2+ resulted in cells with attached parent flagella but detached daughter flagella (Vickerman, 1969). Based on this observation, we developed a simple procedure in which BSF cells were cultured for 18–30 h at 37°C in HMI-9 complete medium supplemented with either 5–20 mM sodium citrate or 2.5–10 mM EGTA to chelate Ca2+, and 1–5 mM MgCl2 (Oberholzer et al., 2011). Under these conditions, many cells had detached flagella, including a subset of 1K1N cells that had a full-length FAZ as demonstrated by L3B2 staining, and had the sole flagellum detached from the cell body except at the flagellar base. Live cells in medium were settled on coverslips for fixation and staining as described below.
Staining for fluorescence microscopy
Except for counts of nuclei/kinetoplast morphology, for all immunofluorescence microscopy, live cells were either (i1) centrifuged at 500–1200 g for 5–14 min to concentrate and diluted in 37°C fresh HMI-9 complete medium or HBS/G (25 mM HEPES pH 7.1, 140 mM NaCl, 5 mM KCl, 0.75 mM Na2HPO4, 22 mM glucose) to 1.5–2.0×106 cells/ml, or (2) cells at 0.8×106–1.2×106 cells/ml were settled directly onto coverslips (#1.5H high precision-type; Paul Marienfeld GmbH & Co. KG) coated with poly-L-lysine (Amanda Polymers, Birmingham, AL) for 20–40 min. For formaldehyde-only fixation, cells on coverslips were washed once or twice in HBS/G, then fixed immediately in HMI-9 complete medium, or HBS/G buffer, at 37°C with freshly added formaldehyde (Ted Pella, Inc.) to 0.75% to 4% final concentration for 10–30 min at room temperature. For formaldehyde plus methanol fixation, cells on coverslips were washed once or twice in HBS/G, then fixed immediately in HMI-9 complete medium, or HBS/G buffer, at 37°C with freshly added formaldehyde (Ted Pella, Inc.) to 0.50% to 1% final concentration for 7–15 min at room temperature. Coverslips were then washed two or three times in HBS/G, placed on ice, and fixed with −20°C methanol for 15–30 min. Fixation was followed by three to five washes in cold PEME (100 mM PIPES pH6.9, 2 mM EGTA, 1 mM Mg2SO4, 0.1 mM EDTA) (Robinson et al., 1991) and a 10–20-min rehydration in cold PEME. For methanol-only fixation (including flagellar length or intensity measurements), cells on coverslips were washed twice in HBS/G, placed on ice and fixed in −20°C methanol for 10–40 min, followed by PEME washing and rehydration steps as above.
When used, detergent extraction was carried out before fixation on live unfixed cells at room temperature using the Nonidet P-40 in PEME buffer as described previously (Robinson et al., 1991), except that (1) live cells were settled onto and extracted on poly-L-lysine-coated coverslips; (2) IGEPAL CA-630 was substituted for Nonidet P-40 at 1% for 5 min at room temperature; and (3) cell extraction buffer and extracted cytoskeletons were stabilized in in HBS/G or PEME/G buffers supplemented with 2× cOmplete™ protease inhibitor cocktail, EDTA-free (Roche Diagnostics), plus 1 mM E 64d cysteine protease inhibitor (Apex Biotechnology, Taiwan). For extracted cells, buffers used prior to fixation were supplemented with 5 mM ATP (molecular biology grade; Millipore Sigma) and 2 mM MgCl2 (except as noted below for cells expressing TbKin2a–Myc).
Following fixation, non-extracted cells were permeabilized (for formaldehyde-only or formaldehyde plus methanol-fixed cells) were permeabilized in PEME plus 0.1% IGEPAL for 5–8 min and washed in PEME buffer. All fixed cells were blocked and incubated in primary then secondary antibodies in PEMEBALG [PEME plus 1% bovine serum albumin (BSA), 0.1% sodium azide, 100 mM lysine, 0.5% cold water fish skin gelatin solution (Millipore Sigma) plus normal goat serum] as described previously (Sagolla et al., 2006) at the following concentrations: L8C4 (Kohl et al., 1999), 1:100; KMX (Birkett et al., 1985), 1:25-1:50; L3B2 mouse anti-FAZ (Kohl et al., 1999), 1:10−1:100; YL1/2 (cat. no. MAB1864-I, MilliporeSigma, Burlington, MA), 1:100; BBA4 mouse monoclonal anti-basal-body (Woods et al., 1989), 1:50; L1C6 mouse anti-nucleolar protein (Durand-Dubief and Bastin, 2003), 1:200; mouse anti-IFT144 (PIFTF6) (Absalon et al., 2008), 1:200; mouse anti-IFT172 (Absalon et al., 2008), 1:1000; rabbit anti-TbKin2a (this study), 1:20,000. Secondary antibodies conjugated to Alexa Fluor 488 or Alexa Fluor 568 (Invitrogen) were diluted at 1:400, with DAPI at 1-10 μg/ml. For double staining with mouse anti-IFT172 and mouse L3B2, anti-IFT172 was labeled with Zenon Alexa Fluor 488 (Invitrogen) and used at a 1:1000 dilution, and L3B2 was labeled using Zenon Alexa Fluor 568 (Invitrogen) and used at 1:10, using the recommended Zenon protocol. Coverslips were mounted onto slides using ProLong Gold antifade and allowed to set hard in the dark at 4°C for at least 24 h, then sealed with clear nail polish before imaging.
For nuclei and kinetoplast morphology counts, 0.7×107–2.0×107 live cells were centrifuged, washed in PBS/G (PBS with 20 mM glucose), then fixed with 4% formaldehyde for 20–30 min. Fixed cells were isolated by centrifugation, washed in PBS, placed on poly-L-lysine-coated coverslips and post-fixed with −20°C methanol for 20–30 min. Fixed cells on coverslips were permeabilized and blocked in PBS with 1% BSA and 0.1% Triton X-100 for 1 h. For staining, cells were incubated with mouse anti-PFR L8C4 antibody (Kohl et al., 1999) at 1:50 in PBS with 1% BSA and then with Alexa Fluor 488 anti-mouse-IgG secondary antibody (Invitrogen) and DAPI (1–10 μg/ml). Samples were mounted using ProLong Gold antifade (Invitrogen) and allowed to set hard in the dark at 4°C for at least 24 h, then sealed with clear nail polish before imaging.
For cells expressing TbKin2a–Myc, transfected cells in medium were placed onto poly-L-lysine-coated coverslips, allowed to settle briefly, and washed in HBS/G. Cells were detergent-extracted using the Nonidet P-40 method in PEME/G as described previously (Robinson et al., 1991) and fixed in −20°C methanol, as noted above. HBS/G and PEME/G buffers were supplemented as noted above except that concentrations of ATP and MgCl2 used were 10 mM and 4 mM, respectively. Cells were blocked and incubated in primary then secondary antibodies as described previously (Sagolla et al., 2006) at the following concentrations: L3B2 mouse IgG1 anti-FAZ monoclonal antibody (Kohl et al., 1999), 1:50–1:100; anti-Myc mouse IgG2a monoclonal antibody 9B11 (Cell Signaling Technology, Danvers, MA), 1:8000. Secondary anti-mouse IgG1 antibodies conjugated to Alexa Fluor 568, and anti-mouse IgG2a antibodies conjugated to Alexa Fluor 488 (Invitrogen), respectively, were diluted at 1:400, and DAPI was used at 1–10 µg/ml. Coverslips were mounted onto slides using ProLong Gold antifade and allowed to set hard in the dark at 4°C for at least 24 h, then sealed with clear nail polish before imaging.
Microscopy and image analysis
Epifluorescence microscopy was performed using an Olympus IX71 inverted microscope with 60× (1.40 NA) and 100× (1.35 NA) PlanApo objectives and a Coolsnap HQ camera (Photometrics, Tucson, AZ) or an Olympus IX71 inverted microscope with UPlanApo 100× (1.35 NA) or UPlanApo 60× (1.40 NA) objectives and Coolsnap ES2 camera (Photometrics). Images were captured using Metamorph (Molecular Devices) or µManager software (Edelstein et al., 2010). Brightness/contrast levels were adjusted using ImageJ/Fiji version 2.0,0-rd-69/1.52p (Schneider et al., 2012) and Photoshop CS6 software (Adobe, San Jose, CA). For flagellar intensity measurements, ImageJ/Fiji was used to measure mean fluorescence intensity at 0.1-mm intervals along 53 individual mature flagella (i.e. data do not include new flagella) on methanol-fixed cells from three independent experiments. For flagellar length measurements, ImageJ/Fiji were used to measure lengths of PFR (stained with L8C4) in uninduced and induced methanol-fixed cells from four independent experiments.
Deconvolution microscopy was carried out using an Applied Precision DeltaVision Elite system (GE Healthcare Life Sciences, Pittsburgh, PA) with an Olympus IX71 inverted microscope with UPlanApo 100× (1.35 NA) or UPlanApo 60× (1.40 NA) objectives and Coolsnap ES2 camera (Photometrics). Images were captured using Deltavision SoftWoRx v6 software (GE Healthcare). From 40–200 serial z-sections per channel were acquired at 0.10–0.40-µm intervals. Image z-stacks were deconvolved using SVI Huygens Professional v14 (Scientific Volume Imaging B.V., Hilversum, Netherlands) using Huygens' Classic Maximum Likelihood Estimation algorithm and experimental point source function (PSF) data generated by the UC Berkeley Biological Imaging Facility (‘deconvolved z-stacks’). Images were analyzed using Bitplane Imaris v7 (Bitplane, Zurich, Switzerland) and ImageJ/Fiji. Two-dimensional maximum intensity projection images were generated from three-dimensional deconvolved stacks using Imaris, ImageJ/Fiji or Huygens Professional software. Image adjustments required for presentation were made using ImageJ/Fiji and Photoshop CS6.
For transmission electron microscopy (TEM), uninduced and induced RNAi cells were processed and thin sections generated as previously described (Tu et al., 2005), and were imaged using a JEOL 1200 transmission electron microscope.
Statistical analyses
Quantitative data were calculated and graphed using Excel (Microsoft Inc., Redmond, WA) and Prism 6 (GraphPad Software, La Jolla, CA). Sample sizes (n) were sufficient to detect statistical significance at the levels indicated. For colocalization analysis of IFT172 and IFT144 with TbKin2a (see Dunn et al., 2011), ImarisColoc (Bitplane, Zurich, Switzerland) was applied to deconvolved image z-stack data cropped to basal bodies-sized and single cell-sized blocks. For all such blocks, ImarisColoc was used to measure total threshold-adjusted signal colocalization for IFT172 and TbKin2a, or IFT144 and TbKin2a channel pairs. Threshold signal intensity lower cutoffs were determined per channel after evaluation of controls and relative sensitivities, and were generally set at 10% of maximum channel intensities. In Fig. 3F, final adjusted Mander's colocalization coefficient and adjusted Pearson's covariance coefficient statistics were computed and graphed using Prism 6. For Fig. 6A, the relative length statistics were computed using one-way analysis of variance (ANOVA) with Sidak's multiple comparison test with variance pooling, in Prism 6. Mean length differences between uninduced versus induced cell lines were significant at P<0.0001 using two-tailed tests. Mean length differences among uninduced cell lines, and between single knockdown induced cell lines, were all non-significant at P<0.05 using two tailed tests. For Fig. 6G, settling data statistics based on A600nm absorption time point measurements were computed with Prism 6 using grouping analyses of uninduced versus induced cells by RNAi cell line using multiple t-tests, one two tailed t-test per cell line pair (induced versus uninduced) per time point (row), corrected for multiple comparisons using the Holm–Sidak method with 0.05 significance levels.
Acknowledgements
We dedicate this paper to the memory of Professor Ching C. Wang (1936-2017) of the University of California, San Francisco. We thank former members of the Wang lab – Ziyin Li, Xiaoming Tu, Stephane Gourguechon, and Praveen Kumar, among others – for reagents, procedures, unpublished observations, draft manuscript review (C.C.W. and Z.L.), insights, and encouragement. We thank Philippe Bastin, Benjamin Morga, and Thierry Blisnick of the Pasteur Institute for antibodies to IFT172 and IFT144, unpublished images, insights, and helpful suggestions. We thank Sam Dean at Oxford University for the generous gifts of pPOTv4 and pPOTv7 plasmids, as well as information, and advice. We thank Scott Dawson at UC Davis for insights and advice on phylogenetic methods and interpretation. We thank Ryan Case for insights and suggestions regarding bioinformatic analysis of kinesin motor domains and neck linkers. For reagents we thank: Keith Gull at Oxford University for antibodies (KMX, L8C4, L1B6, L3B2 and BBA4), Paul Englund at Johns Hopkins University for the pZJM vector, and George Cross at Rockefeller University for the 90-13 cell line. Denise Schichnes and Steve Ruzin of the UC Berkeley Biological Imaging Facility provided assistance with deconvolution microscopy, image processing, and analysis. Finally, we thank Welch lab members for their suggestions, support and patience.
Footnotes
Author contributions
Conceptualization: R.L.D., B.M.H., A.A., M.D.W.; Methodology: R.L.D., B.M.H., A.A., H.W., R.L.J., J.M., W.Z.C., M.D.W.; Validation: R.L.D., B.M.H., A.A., H.W., R.L.J.; Formal analysis: R.L.D., B.M.H., A.A., H.W., M.D.W.; Investigation: R.L.D., B.M.H., A.A., H.W., R.L.J., J.M.; Data curation: R.L.D., B.M.H., A.A., H.W., W.Z.C., M.D.W.; Writing - original draft: R.L.D., M.D.W.; Writing - review & editing: R.L.D., B.M.H., A.A., H.W., R.L.J., J.M., W.Z.C., M.D.W.; Visualization: R.L.D., B.M.H., A.A., H.W., J.M.; Supervision: M.D.W.; Project administration: M.D.W.; Funding acquisition: R.L.D., M.D.W.
Funding
This work was funded in part by grant A40732 from the UNDP/World Bank/WHO Special Programme for Research and Training In Tropical Diseases (TDR). Funding was also provided by R.L.D., including through a family trust.
References
Competing interests
The authors declare no competing or financial interests.