Flagellar Motility Contributes to Cytokinesis in Trypanosoma brucei and Is Modulated by an Evolutionarily Conserved Dynein Regulatory System

Procyclic culture form (PCF) 29-13 cells, which stably express T7 RNA polymerase and tetracycline (Tet) repressor (74), were used to generate all RNAi knockdown mutants. Trypanosomes were cultivated and stably transfected as described previously (20, 26). Clonal cell lines were obtained by limiting dilution. Clonal lines harboring the RNAi constructs p2T7^-Ti-B/RSP3, p2T7^-Ti-B/PF16, p2T7^-Ti-B/PF20, p2T7^-Ti-B/PFR2-1, p2T7^-Ti-B/PFR2-2, p2T7^-Ti-B/TPN, p2T7^-Ti-B/RSP3-TPN, or p2T7^-Ti-B/PF16-TPN (see section on DNA constructs, below) are referred to as TbRSP3, TbPF16, TbPF20, TbPFR2-1, TbPFR2-2, TbTPN, TbRSP3/TPN, and TbPF16/TPN, respectively. For tetracycline induction, cells were split into two flasks and cultured with or without 1 μg of tetracycline/ml and diluted as necessary to maintain exponential growth. For growth curves, cell densities were measured by using a hemacytometer, and the averages of two independent counts are reported. For knockdown mutants, clusters of up to four distinguishable cell bodies were counted. To culture cells with physical shaking, TbRSP3 cells were split into flasks with or without tetracycline and placed on an orbital shaker set at 80 rpm.

C. reinhardtii protein sequences for RSP3 (accession number P12759) (73), PF16 (accession number AAC49169.1) (59), and PF20 (accession number AAB41727.1) (60) were used to identify homologues in the T. brucei genome database (http://www.sanger.ac.uk/cgi-bin/BLAST/submitblast/t_brucei/omni ) by using WU-BLAST (Washington University, St. Louis, MI). Sequencing of the T. brucei genome was accomplished as part of the Trypanosoma Genome Network with support by The Wellcome Trust. We identified single-copy genes for RSP3 (GeneDB ID Tb11.47.0034), PF16 (GeneDB ID Tb927.1.2670), and PF20 (GeneDB ID Tb10.61.2920). The percent similarity of each T. brucei protein to its C. reinhardtii homologue is as follows: RSP3 (41.5% identity and 52.5% similarity), PF16 (55.9% identity and 69.2% similarity), and PF20 (37.3% identity and 50.6% similarity).

Multiple sequence alignments were performed with the CLUSTAL W algorithm (65) by using Vector NTI software (Vector NTI Advance Suite 8; InforMax, Inc., Bethesda, MD). For RSP3, the following proteins were aligned: Homo sapiens (accession number AAK26432), C. reinhardtii (accession number P12759), T. brucei (GeneDB ID Tb11.47.0034), T. cruzi (GeneDB ID Tc00.1047053506721.20), and Leishmania major (GeneDB ID LmjF27.0520). For PF16, the following proteins were aligned: H. sapiens (accession number CAH72210), C. reinhardtii (accession number AAC49169.1), T. brucei (GeneDB ID Tb927.1.2670), T. cruzi (GeneDB ID Tc00.1047053510955.40), and L. major (GeneDB ID LmjF20.1400). For PF20, the following proteins were aligned: H. sapiens (accession number AAM63956), C. reinhardtii (accession number AAB41727.1), T. brucei (GeneDB ID Tb10.61.2920), T. cruzi (GeneDB Tc00.1047053511277.340), and L. major (GeneDB ID LmjF18.0470).

All RNAi plasmids were constructed in p2T7^-Ti-B, which contains opposing, tetracycline-inducible T7 promoters such that tetracycline-induced transcription generates an intermolecular double-stranded RNA (dsRNA) (30). To create p2T7^-Ti-B/RSP3, a 375-bp fragment corresponding to nucleotides (nt) 566 to 940 of the T. brucei RSP3 open reading frame (ORF) was PCR amplified from 29-13 genomic DNA and cloned in the forward orientation into the HindIII and SacII sites in p2T7^-Ti-B. To create p2T7^-Ti-B/PF16, a 247-bp fragment corresponding to nt 1288 to 1534 of the T. brucei PF16 ORF was PCR amplified from 29-13 genomic DNA and cloned in the forward orientation into the SacII and XbaI sites in p2T7^-Ti-B. To create p2T7^-Ti-B/PF20, a 407-bp fragment corresponding to nt 291 to 697 of the T. brucei PF20 ORF was PCR amplified from 29-13 genomic DNA and cloned in the forward orientation into the SacII and XbaI sites in p2T7^-Ti-B. To create p2T7^-Ti-B/PFR2-1, a 448-bp fragment corresponding to nt 642 to 1089 of the T. brucei PFR2 ORF (GeneDB ID Tb08.5H5920) was PCR amplified from 29-13 genomic DNA and cloned in the forward orientation into the HindIII and BamHI sites in p2T7^-Ti-B. Note that this fragment of PFR2 was selected by RNAit (49) and is not expected to target PFR1 (GeneDB ID Tb03.26J7510) mRNA, since it does not contain any regions of identity to PFR1 longer than 12 nt. To create p2T7^-Ti-B/PFR2-2, a 1,723-bp fragment corresponding to nt 78 to 1800 of the T. brucei PFR2 ORF was PCR amplified from 29-13 genomic DNA and cloned in the forward orientation into the HindIII and BamHI sites in p2T7^-Ti-B. This fragment of PFR2 is identical to the fragment used to create the snl-2 cell line (4) and is similar to the fragments used to create snl-1 (nt 78 to 1758) (7) and PFRAi (nt 78 to 1779) (14). To create p2T7^-Ti-B/TPN, a 252-bp fragment corresponding to nt 1103 to 1354 of the T. brucei trypanin ORF (26) was PCR amplified from 29-13 genomic DNA and cloned in the forward orientation into the BamHI and HindIII sites in p2T7^-Ti-B. All RNAi constructs were verified by restriction digestion and direct sequencing. Plasmids were linearized at the unique EcoRV or NotI site for transfection.

To create dual RNAi knockdown plasmids, tandem fragments of the targeted genes were cloned into p2T7^-Ti-B. To construct p2T7^-Ti-B/RSP3-TPN, the precise 375-bp RSP3 fragment that was used to create p2T7^-Ti-B/RSP3 was cloned in the forward orientation into the XbaI and BamHI sites, upstream of the TPN fragment in p2T7^-Ti-B/TPN. To create p2T7^-Ti-B/PF16-TPN, the precise 247-bp PF16 fragment that was used to create p2T7^-Ti-B/PF16 was cloned in the forward orientation into the XbaI and BamHI sites, upstream of the TPN fragment in p2T7^-Ti-B/TPN.

Total RNA samples were prepared by using an RNeasy kit (QIAGEN) according to the manufacturer's instructions. RNA samples (5 μg) were analyzed by Northern blotting as described previously (21). ³²P-labeled probes corresponding to nt 198 to 546 of the RSP3 ORF (RSP3 probe), nt 1288 to 1534 of the PF16 ORF (PF16 probe), nt 291 to 697 of the PF20 ORF (PF20 probe), nt 642 to 1089 of the PFR2 ORF (PFR2 probe), nt 1 to 775 of the TPN ORF (TPN probe), or the entire TRP ORF (TRP probe) were used. The TRP gene (GeneDB ID Tb09.244.2800) encodes a “trypanin-related protein” with 28% identity and 43% similarity to trypanin and will be described elsewhere (K. L. Hill, unpublished observation).

Sedimentation assays were performed as described previously (6). Briefly, cells were incubated with or without tetracycline for 24 h and then resuspended to 5 × 10⁶ cells/ml in fresh medium. Each culture was divided into aliquots to four cuvettes (1 ml per cuvette) and incubated under standard growth conditions. The optical density at 600 nm (OD₆₀₀) was measured every 2 h. At each time point, two cuvettes from each culture were left undisturbed to monitor sedimentation, while the other two cuvettes were resuspended to monitor growth. The ΔOD₆₀₀ for each sample was calculated by subtracting the OD₆₀₀ of the resuspended samples from that of the undisturbed samples.

Protein extracts were prepared and analyzed by Western blotting (20). Trypanin was detected with a monoclonal antibody directed against a synthetic peptide corresponding to the last 13 amino acids of trypanin and was generated by an outside vendor (Cell Essentials, Cambridge, MA). PFR2 was detected with the monoclonal antibody L8C4 (29). The monoclonal antibody E7, directed against β-tubulin, was used as a control for protein loading. This antibody was developed (10) and obtained from the Developmental Studies Hybridoma Bank maintained by the University of Iowa Department of Biological Sciences.

Live cells were imaged by using a Zeiss Axiovert 200 M inverted microscope with a 63× Achroplan LD or 63× Plan-Neofluor oil-immersion objective. Video images were captured as described below. For fluorescence microscopy, cells were imaged on a Zeiss Axioskop II compound microscope with a 63× Plan-Neofluor oil immersion objective, and images were captured by using a Zeiss Axiocam digital camera and Zeiss Axiovision 3.0 software.

TbRSP3, TbPF16, and TbPF20 cells were grown in the presence or absence of 1 μg of tetracycline/ml (96 h for TbRSP3 or 60 h for TbPF16 and TbPF20). Cytoskeletons or whole cells were prepared (19, 51) and then fixed for 60 min as described previously (26) with the addition of 1% (wt/vol) tannic acid. Samples were then washed in fixative without tannic acid and shipped overnight to Yale University, where processing was completed as described previously (58), except that staining en bloc was done in 2% uranyl acetate for 2 h.

Cytoskeletons were used for radial spoke analysis. Since a full complement of nine radial spokes was not always visible in all control sections, we established a blind assay to compare tetracycline-induced rsp3 mutants to uninduced controls. Random silver sections from control (−Tet, n = 32) and tetracycline-induced (+Tet, n = 31) samples were scanned for axonemes in which the central pair microtubules were cut in cross section. These axonemes were photographed, and the images were coded and then mixed and scored blindly to determine the number of spokes present per axoneme. After scoring, samples were decoded, and divided into control and tetracycline-induced groups, and the data are reported as the number of sections in each group having zero to nine spokes per section.

Whole-cell samples were used to measure the orientation of central pair microtubules in control cells and pf16 and pf20 mutants. Samples were selected randomly in which at least four outer doublet microtubules were clearly resolved. To measure central pair orientation, a reference line was drawn from the center point of outer doublet one to the center point of the outer doublet seven. The outer doublet seven can be readily identified by its attachment to the PFR (5). A line was then drawn that bisects the C1 and C2 central pair microtubules, and the angle of intersection of this line with the reference line was measured, moving clockwise from the reference line. The median angle in 45 control cells was 72°. For reference, this angle (72°) was set to zero, and the data were plotted as degrees deflection from zero.

TbRSP3 cells were incubated with or without tetracycline for 26 h and then subjected to a dye exclusion assay for cell viability (Molecular Probes Live/Dead Assay #L-3224) according to the manufacturer's instructions. Briefly, cells were washed once and resuspended to 2 × 10⁷ cells/ml in 1× phosphate-buffered saline, and a 50-μl portion of the cells was then added to uncoated glass coverslips and incubated in 5 μM ethidium homodimer-1 in 1× phosphate-buffered saline for 30 min. Coverslips were inverted onto glass slides, sealed with nail polish, and imaged.

To monitor cell cycle progression, cytoskeletons were prepared by detergent extraction (51) and subjected to immunofluorescence as described previously (26) with the α-PFR2 monoclonal antibody L8C4 (29). Anti-mouse secondary antibodies conjugated to Alexa-Fluor 488 (Molecular Probes) were used at a 1:400 dilution. Samples were mounted in Vectashield H-1200 (Vector Labs) containing 1.5 μg of DAPI (4′,6′-diamidino-2-phenylindole)/ml to visualize kinetoplast and nuclear DNA and then imaged. Cell cycle analysis was performed as described previously (58). Cells were distinguished as having one or two flagella (1F/2F), one or two kinetoplasts (1K/2K), and one nucleus or two discrete nuclei (1N/2N). A separate category (1mN) was used to distinguish cells with mitotic nuclei that contained a visible spindle. To minimize the influence of downstream defects, cells were examined 30 to 48 h postinduction, which corresponded to the onset of the cytokinesis defect.

Live TbRSP3 cells, grown in the presence (+Tet, n = 792) or absence (−Tet, n = 676) of 1 μg of tetracycline/ml for 48 h were analyzed at the same cell densities in logarithmic growth, and two separate samples were quantified for each culture (+Tet, n = 393 and n = 399; −Tet, n = 334 and n = 342), with the standard deviation indicated by error bars. Note that the number of cells in cytokinesis was much higher in live cultures than in cells processed for immunofluorescence or harvested for routine cell counting. This indicated that cells undergoing cytokinesis were easily separated by physical manipulation. To account for this, live cultures were directly examined in culture flasks with minimal physical manipulation. Cells were categorized as having a single cell body or multiple physically attached cell bodies. Cells in which two cell bodies remain attached to one another (“double” in Fig. 4B) were further subdivided into three categories. “Double-1” corresponded to cells that had two full-length flagella but had not visibly begun cytokinesis (stages VI to VIII, as defined in Fig. 31 of reference 58). “Double-2” corresponded to cells that were in the middle-to-late stages of cytokinesis (stage IX in reference 58), while “double-3” corresponded to cells in the final stage of cytokinesis that were attached only by their extreme posterior ends (stage X in reference 58). Note that cells in each of these stages were also present in wild-type cultures (Fig. 4B) (58) and are therefore normal events in the T. brucei cell division cycle. The data shown are representative of three independent experiments.

Videos are accessible in the supplemental material. pf16 single and pf16/tpn double mutants (TbPF16 and TbPF16/TPN cells, respectively) were examined 48 h postinduction by using differential interference contrast (DIC) optics. Cells were visualized on standard glass slides, with double-sided tape between the slides and coverslips creating a chambered space to allow freedom of movement. Slides were inverted and examined through the coverslip. Videos were imaged by using a COHU charge-coupled device analog video camera and converted to digital format with a HandyCam (Sony, Inc.). Digital clips were captured at 30 frames per s and converted to AVI movies with Adobe Premiere Elements software (Adobe Systems). Representative examples are shown in in the supplemental material to illustrate the major phenotypes of each mutant. In Fig. 4A, images shown are single frames from videos S6 to S7 in the supplemental material. In Fig. 5C, the images shown are single frames from videos S9 and S10 in the supplemental material. Figure 8A and B are frames from videos S3 and S11 in the supplemental material, respectively. The frames were rotated to orient posterior end of the cell body upward.

DIC video of live pf16 single and pf16/tpn double mutants (TbPF16 and TbPF16/TPN cells, respectively) was captured 48 h postinduction as described above. For each cell line, footage of 100 random cells was recorded and analyzed. Since the flagellar beat in T. brucei is three-dimensional and bending can occur independently in the distal and proximal portions of the flagellum, it is difficult to reliably measure the frequency and amplitude of a single waveform along the length of the entire flagellum. Therefore, we measured the number of beats in each mutant over a 10-s period, focusing on the flagellar tip, where single mutants are most active. Beats of the flagellar tip were quantified over a 10-s time period for each cell. Where cells exhibited more than 12 beats in 10 s, individual flagellar movements could not be quantified, and these cells were categorized as “>12” beats/10 s, which is essentially continuous movement. An example of a pf16 mutant that had >12 movements in this assay is shown in video S4 in the supplemental material to illustrate the qualitative differences between pf16 and pf16/tpn mutants in the >12 category (see Fig. 8C).

We used RNAi to target radial spokes and the central pair apparatus of T. brucei. To target radial spokes, we chose RSP3, which is required for attachment of the spoke to the axoneme in C. reinhardtii (12). The major components of the central pair apparatus are two singlet microtubules, C1 and C2, together with their specific projections and intermicrotubule bridges (62). To target the central pair apparatus, we chose PF16 and PF20, which are localized to the C1 and C2 microtubule structures, respectively (59, 60) and are required for stability of central pair microtubules in C. reinhardtii (1, 15) and mice (56, 76).

BLAST searches of the T. brucei genome database identified single-copy RSP3, PF16 and PF20 homologues (see Materials and Methods). Protein alignments demonstrated that these proteins are highly conserved, with PF20 exhibiting the most divergence (Fig. 1). In the region of overlap, the percent identity among RSP3, PF16, and PF20 homologues are 27.9, 42.4, and 16.7%, respectively. The C. reinhardtii and human RSP3 proteins have unique C-terminal extensions of 201 and 90 amino acids, respectively, whereas the C. reinhardtii PF16 protein has a unique C-terminal extension of 52 amino acids, suggesting some aspects of their function may be organism specific. Notably, a shorter C terminus is also observed in RSP3 and PF16 from other kinetoplastids (Fig. 1A and B), and deletion studies in C. reinhardtii indicate that the C-terminal 140 amino acids of RSP3 are not required for motility (13).

Separate plasmids for tetracycline-inducible RNAi knockdown of RSP3, PF16, and PF20 were constructed, and stably transfected clonal cell lines were obtained by limiting dilution. These cell lines will be henceforth referred to as TbRSP3, TbPF16, and TbPF20 cells or interchangeably as rsp3, pf16, and pf20 knockdown mutants. Northern blot analysis demonstrates that the targeted transcripts are dramatically reduced within 24 h of tetracycline induction (Fig. 2A). Likewise, visual inspection of rsp3, pf16, and pf20 knockdown mutants revealed a severe motility defect within 24 h of induction (see below). To quantitate this defect, we performed sedimentation assays (Fig. 2B). All three mutants sedimented at a linear rate of approximately −0.03 OD₆₀₀ units/hour, whereas uninduced cells remained suspended. A dye-based viability assay demonstrated that there was no difference in viability at the time of the assay (see Materials and Methods).

Close examination revealed that rsp3, pf16, and pf20 knockdown mutants were immotile in the sense that they exhibited no net movement in any direction, nor did they rotate or tumble as observed for trypanin knockdown mutants (26). However, a rudimentary flagellar beat or twitching was observed in all three mutants, so these cells should be considered “immotile” but not “paralyzed.” This was particularly evident in rsp3 mutants, where a few cells were completely paralyzed but most were able to sustain a rudimentary flagellar beat (video S2 in the supplemental material). Nonetheless, flagellar movement was generally confined to the distal portion of the flagellum, was mostly planar, and did not drive rotation of the cell body. Electron microscopy demonstrated that the number of spokes was reduced in axonemes of rsp3 mutants compared to controls (Fig. 2C and D). In control cells, six or more spokes were clearly visible in 74% of sections (n = 32) compared to only 25% of sections from rsp3 mutants (n = 31). The average number of spokes per section was reduced to 4.2 in rsp3 mutants compared to 6.6 in controls.

pf16 and pf20 mutants were also immotile and incapable of cellular rotation (videos S3 and S5 in the supplemental material). However, the beat defect was more severe than in rsp3 mutants (compare videos S3 and S5 with video S2 in the supplemental material). Whereas rsp3 mutants were able to sustain a modest flagellar beat, the pf16 and pf20 mutants only twitched erratically, and this irregular twitching was confined to the distal region of the flagellum. An additional feature shared by pf16 and pf20 mutants, but not rsp3 mutants, was an exaggerated curvature in the cell body such that the flagellum tip and the anterior end of the cell were bent in an “S” shape toward the side of the cell opposite the flagellum (videos S3 and S5 in the supplemental material and Fig. 8A). In these cells, the flagellum was situated along the outer edge of the bend, such that flagellar movement decreased the degree of bending but was not powerful enough to fully straighten the cell body.

Electron microscopy of pf16 and pf20 mutants (Fig. 3) demonstrated a striking alteration in the orientation of the central pair microtubules. In control cells, central pair orientation within the axoneme was found to be restricted to a very narrow range, such that a plane bisecting the C1 and C2 microtubules was always roughly parallel to the paraflagellar rod viewed in cross-section (Fig. 3A and C). In contrast, orientation of the central pair was highly variable in pf16 and pf20 mutants (Fig. 3B and C). Since we cannot distinguish between the two central pair microtubules, we do not know whether this represents random orientation through 360° or only 180°.

RNAi knockdown of RSP3, PF16, and PF20 resulted in a severe motility defect that was evident within 24 h of tetracycline induction. Surprisingly, in all three mutants the loss of motility was invariably followed by a cell division defect such that cells failed to complete normal cell division and accumulated as clusters attached at their extreme posterior ends (Fig. 4A, and videos S6 to S7 in the supplemental material). Cleavage furrow formation in T. brucei initiates at the anterior end of the cell, between the tips of the new and old flagella, and then proceeds toward the posterior end, and in the final stages of cytokinesis daughter cells are only connected at their posterior ends (58). In wild-type cells, these daughter cells eventually pull apart before the next round of cytokinesis, while the motility mutants do not and therefore accumulate as larger and larger clusters. This defect was investigated more thoroughly in rsp3 mutants, where reduced severity of flagellar paralysis allowed us to examine early aspects of the phenotype. Within 48 h postinduction, there was a decrease in the number of single cells and an increase in cell clusters containing two, three, four, and more than four conjoined cell bodies. Among the “double” cells (as defined in Materials and Methods and Fig. 4), there was no significant change in cells that had not initiated cytokinesis (Fig. 4B, double-1), whereas there was a significant increase in cells in the middle to late stages of cytokinesis (Fig. 4B, double-2). There was a modest decrease in the number of double-3 cells, perhaps indicating that cells undergoing cytokinesis must fully complete the double-2 stage before reaching the final stage. Analysis of cell cycle progression by immunofluorescence (see Materials and Methods) demonstrated that, aside from cells blocked in cytokinesis, there was no accumulation at any other stage of the cell cycle (not shown). These data, together with the fact that new rounds of cytokinesis can be initiated in “double” cells to produce “triple” and “quad” clusters, indicated that initiation of cytokinesis was not significantly affected but that daughter cells were unable to complete cell separation.

Knockdown of three different axonemal proteins resulted in loss of cell motility and a concomitant cytokinesis defect. Since the cytokinesis defect was unexpected, we sought to determine whether a similar defect would be observed when we targeted a nonaxonemal protein that is required for normal cell motility. PFR2 is one of two major proteins that comprise the kinetoplastid paraflagellar rod (5). Previously described PFR2 knockdown mutants had a reduced flagellar beat and were immotile at the cellular level (4, 7). These mutants exhibited a reduced growth rate compared to the wild type (6), although no specific cytokinesis defect was reported. To further investigate the connection between flagellar beat and cytokinesis, we generated two independent, tetracycline-inducible PFR2 knockdown mutants. Tetracycline induction led to a rapid and dramatic loss of PFR2 mRNA and protein in both mutants (Fig. 5A and B). The flagella of these mutants were able to beat, but beating was not sufficient to drive directional cell motility or significant cellular rotation (video S8 in the supplemental material). Within 24 h of tetracycline induction, pfr2 mutants began to accumulate as small clusters of cells attached at their posterior ends (Fig. 5C and videos S9 and S10 in the supplemental material), and a reduction in growth rate was observed (Fig. 5E). By 4 days postinduction, the majority of cells accumulated in massive clusters (Fig. 5D). Therefore, as was the case for central pair and radial spoke proteins, loss of PFR2 led to reduced flagellar beat, a loss of cell motility, and a concomitant block in cytokinesis.

Failure to complete cytokinesis in four independent motility mutants might reflect an active requirement for flagellar beat. Alternatively, the effect might be indirect, with an impaired beat causing internal structural defects that prevent the separation of daughter cells. If flagellar beat contributes directly to cell separation, gentle agitation of the mutant cultures may provide compensatory forces that allow cell division. On the other hand, if the effect is simply indirect, i.e., if flagellar paralysis simply distorts something inside the cell and as a result cells are inextricably intertwined, agitation is not expected to correct these defects and clusters should still form. Growth rate and cytokinesis were examined in motility mutants maintained normally (“unshaken”) or on a rotating platform (“shaken”). Agitation of the culture completely rescued the cytokinesis defect, restored the growth rate (Fig. 6A), and prevented the formation of multicellular clusters (Fig. 6B).

The motivation for generating radial spoke and central pair mutants was to determine whether a dynein regulatory system provides communication between CP/RS and axonemal dynein in T. brucei. The defining feature of DRC genes is that loss-of-function mutations in these genes suppress flagellar beat defects of central pair and radial spoke loss-of-function mutants (24). Tetracycline-inducible pf16/tpn and rsp3/tpn dual RNAi knockdown mutants were generated as described in Materials and Methods. Northern blot (Fig. 7A and B) and Western blot (Fig. 7C) analyses demonstrated that knockdown in double mutants was as efficient as knockdown in the corresponding single mutants.

The motility defect of rsp3 and pf16 single mutants in T. brucei is defined by three shared phenotypic characteristics: (i) sedimentation, (ii) an abnormal flagellar beat that is often restricted to the flagellar tip, and (iii) cytokinesis failure. In addition, flagellar beat in pf16 mutants is restricted to an erratic twitching, resulting in a sharp curvature in the anterior cell body. The ability of trypanin knockdown to suppress any or all of these phenotypic characteristics was determined. Loss of trypanin did not suppress overall cellular movement defects, since the growth rate and sedimentation profiles of double mutants were similar to single mutants (not shown). However, the key feature of DRC suppression is improved flagellar beat, which was therefore examined with high-resolution phase contrast and DIC microscopy. Although the less-severe phenotype of rsp3 mutants made it difficult to reliably detect improved flagellar beat, suppression of the pf16 beat defect was immediately obvious in pf16/tpn double knockdown mutants (Fig. 8A and B, and videos S11 and S12 in the supplemental material). In contrast to the erratic flagellar twitch exhibited by pf16 single mutants, pf16/tpn double mutants produced a sustained flagellar waveform and rarely twitched (compare videos S11 and S12 to video S3 in the supplemental material). Likewise, the curvature of the anterior cell body and flagellum, which is a hallmark of pf16 single mutants, was reduced or absent in double mutants (Fig. 8A and B), most likely as a direct result of the more regular and controlled beating of the flagellum.

To demonstrate the suppression afforded by loss of trypanin, the phenotypic characteristics of these mutants were used to distinguish pf16 single and pf16/tpn double mutants in a blind assay. Fifty separate samples of pf16 and pf16/tpn mutants were visually inspected and mutants were accurately identified as single (pf16) or double (pf16/tpn) (Table 1). We also performed a quantitative analysis of flagellar motion to distinguish single and double mutants (Fig. 8C). A total of 100 pf16 single mutants and 100 pf16/tpn double mutants were examined, and this analysis showed that flagella of most double mutants beat continuously, whereas single mutants do not (Fig. 8C). Therefore, loss of trypanin suppresses the erratic twitching and curved flagellum that are hallmarks of pf16 mutants.

Many cellular functions are dependent upon correct spatial and temporal regulation of dynein motors, and identification of the proteins and mechanisms underlying this regulation represents a major challenge in cell biology. In C. reinhardtii, the DRC functions as part of a mechanochemical signal transduction pathway that regulates the axonemal dynein in response to signals from central pair microtubules and radial spokes (24, 43, 44, 55). DRC components are defined through their genetic interaction with central pair components. Specifically, loss-of-function DRC mutations suppress flagellar beat defects of central pair loss-of-function mutants (24). Using dual RNAi knockdown, we show that the loss of trypanin suppresses the flagellar beat defects of a central pair (pf16) knockdown mutant in T. brucei, demonstrating a genetic interaction between trypanin and PF16. Therefore, trypanin meets the following criteria of a DRC component: (i) trypanin loss suppresses a central pair beat defect without restoring expression of the central pair protein, (ii) trypanin single knockdown mutants exhibit a motility defect indicative of abnormal dynein regulation (26), (iii) trypanin is stably associated with flagellar axonemes (19), and (iv) is distributed along the length of the flagellum (26). Taken together, these data show that trypanin functions as part of a DRC-like regulatory system in African trypanosomes. Although we did not readily detect a genetic interaction between RSP3 and trypanin, this is probably because the less severe rsp3 phenotype makes suppression too subtle to detect.

Trypanin single-knockdown mutants are incapable of directional cell motility but have an actively beating flagellum and are able to spin and tumble in place (26). Interestingly, a tumbling motion similar to the trypanin knockdown phenotype is also observed transiently in wild-type cells since they alternate between directional runs and random tumbles (18, 26). Therefore, the DRC of trypanosomes may provide a clutch-like regulatory mechanism that can be engaged for directional movement or disengaged for random tumbling. In bacteria, this type of behavior is exploited to drive changes in taxis in response to environmental cues (33).

A DRC function for trypanin is also supported by the localization of trypanin along the flagellum and may explain the flagellum attachment defect of trypanin knockdown mutants. The T. brucei flagellum is directly connected to the cell body along its length and trypanin knockdown mutants exhibit partial flagellar detachment (26). This defect is more pronounced upon the removal of cellular membranes, suggesting that the loss of trypanin compromises the structural integrity of flagellum attachment complexes (26). Since the DRC functions in beat regulation, flagellum detachment in tpn knockdown mutants might be an indirect consequence of the motility defect, as previously suggested (55), rather than a direct consequence of the loss of trypanin. These two explanations are not mutually exclusive, and a third possibility is that trypanin functions in both capacities. The 96-nm periodicity of the DRC along the axoneme in C. reinhardtii (16) is the same as the periodicity of the flagellum attachment complexes in T. brucei (67). Therefore, the DRC or an associated structure might be exploited to serve a secondary function as an axonemal attachment site. Distinguishing between these possibilities will require more precise ultrastructural localization of trypanin and the DRC in T. brucei.

Further analysis of flagellar dynein regulation in T. brucei will be of tremendous interest, particularly since the trypanosome genome also contains a trypanin paralogue, TRP (for trypanin-related protein [see Materials and Methods]). TRP mRNA is not targeted by trypanin RNAi (Fig. 7A) and the contribution of TRP to dynein regulation remains to be determined. Interestingly, genes for both trypanin and TRP are found in other kinetoplastids, whereas C. reinhardtii and humans have only a single trypanin homologue (K. L. Hill, unpublished observation). The presence of a paraflagellar rod in kinetoplastids imparts structural asymmetry to the axoneme, since outer doublets four through seven are physically linked to the paraflagellar rod (5). This asymmetric arrangement undoubtedly imposes unique regulatory demands, and perhaps this need is met with two distinct dynein regulatory systems, one containing trypanin and the other containing TRP.

We identified T. brucei RSP3, PF16, and PF20 homologues as single-copy genes in the T. brucei genome. Knockdown of each of these genes results in severely defective motility. Ultrastructural analysis demonstrates loss of radial spokes in rsp3 mutants and abnormal central pair orientation in pf16 and pf20 mutants. In vitro studies with reactivated axonemes show that the position of the C1 microtubule of the central pair correlates with the position of active dyneins in C. reinhardtii (71, 72). Hence, the restricted orientation of central pair microtubules, together with the aberrant central pair orientation in pf16 and pf20 mutants, supports the idea that all dyneins around the circumference of the axoneme do not receive the same regulatory inputs in T. brucei. It is possible that the abnormal central pair orientation in T. brucei pf16 and pf20 mutants leads to aberrant dynein activity and abnormal motility, but currently we do not know whether abnormal central pair orientation is the cause or the consequence of the motility defect. We did not observe a loss of central pair microtubules in intact cells (Fig. 3) or demembranated flagella (not shown) from pf16 or pf20 mutants. In contrast, pf16 and pf20 deficiency in C. reinhardtii (1, 15) and mice (56, 77) manifests as instability of central pair microtubules, although the precise function of these proteins is not known. Interestingly, loss of central pair microtubules in C. reinhardtii pf16 mutants is observed in demembranated axonemes but not in intact cells (15), indicating that something other than the absence of the central pair is responsible for the motility defect. By demonstrating that the presence of central pair microtubules is not sufficient for motility in T. brucei, our results also extend the findings of McKean et al. (35), who demonstrated that complete ablation of the central pair apparatus in T. brucei by gamma tubulin knockdown disrupts cell motility.

A rudimentary beat is often observed in the flagellar tips of rsp3, pf16, and pf20 mutants. Since the flagellar beat in T. brucei initiates at the tip (69, 70), it thus appears that these mutants are able to initiate but not propagate flagellar beat. The pronounced curvature of pf16 and pf20 mutants likely also results from the inability to sustain flagellar beat, since it is reduced or absent in pf16/tpn double mutants and rsp3 single mutants. The twitching and curved flagella seen in T. brucei pf16 and pf20 mutants is similar to the twitching and curved flagella of sperm from pf16 knockout mice (56) and pf20 chimeric mice (76). Twitching and phenotypic heterogeneity is also observed in C. reinhardtii pf16 and pf20 loss-of-function mutants (1, 15), suggesting that residual flagellar movement is independent of PF16 and PF20 and is not simply due to incomplete knockdown.

Radial spokes are composed of an estimated 23 proteins (75), with a surprisingly large number containing predicted regulatory domains, supporting the emerging concept that spokes play important roles in signal transduction and are not simply structural components of the flagellum (62). The central pair apparatus is also composed of at least 23 proteins (1, 15). By establishing a requirement for radial spoke and central pair proteins in trypanosome motility, our results facilitate structure-function analyses by aiding in rational selection of key amino acids to target with site-directed mutagenesis. The repertoire of tools available for molecular genetic manipulation in T. brucei, including targeted gene knockouts and inducible RNAi knockdowns, makes these parasites a powerful system for elucidating the function of these and other (3, 31) candidate motility genes.

An unanticipated finding from our studies is that flagellar motility is required for normal cell division. rsp3, pf16, pf20, and pfr2 knockdown mutants are each blocked at the final stage of cell separation but can reinitiate cytokinesis multiple times. The loss of RSP3, PF16, PF20, or PFR2 affects distinct substructures within the flagellum, and in all four mutants defective flagellar motility precedes the block in cytokinesis. Therefore, flagellar motility itself, rather than a specific flagellar substructure, is required for cytokinesis.

Other motility mutants have been described in T. brucei. Since those that are nonviable are also defective in critical structures such as the mitotic spindle (35) or flagellum attachment zone (28, 30), it is difficult to assess the relative contribution of flagellar beat to the lethal phenotype. PFR2 knockdown mutants, snl-1 (7) and snl-2 (4), have been described previously that exhibit severely reduced flagellar beat. These mutants are viable but grow more slowly than the wild type (6). Cytokinesis was not specifically examined in the previous studies, although one report indicates that snl-2 cells divide normally (4). Reexamination of these mutants demonstrates they form multicellular clusters, albeit not to the extent seen in the pfr2 mutants obtained in the current study (P. Bastin, personal communication). The reason for the difference in severity is not entirely clear, but both snl-1 (7) and snl-2 (4) were isolated under conditions in which PFR2 dsRNA is expressed. Hence, there is selective pressure for compensation to allow outgrowth of these mutants, and the difference in severity may be due to differences in the extent of flagellar beating and/or other compensating factors. Consistent with this explanation, transfection of the snl-2 plasmid into a cell line that lacks a tetracycline repressor and therefore expresses high levels of the PFR2 dsRNA gave poor transfection efficiency, and the few transfectants obtained had severe cytokinesis defects (P. Bastin, personal communication). An independent pfr2 mutant generated utilizing an intermolecular dsRNA is viable (14) but also forms multicellular clusters (P. Bastin, personal communication). Finally, attempts to generate a PFR2 knockout via gene disruption in T. brucei have failed (25), a finding consistent with the idea that the PFR2 gene is essential. Therefore, the combined data support a requirement for flagellar motility in cytokinesis in T. brucei.

The cytokinesis defect of motility mutants might reflect a direct or indirect role for flagellar beat or a combination of direct and indirect roles. Several lines of evidence suggest that a direct role is at least partially responsible. The initial defect is marked by an accumulation of cells connected at their extreme posterior ends. Hence, cleavage furrow ingression and progression along the flagellum attachment zone is normal, and cell separation fails at the posterior end. Indeed, daughter cells are sometimes connected by only a thin string of membrane (data not shown). It is difficult to envision internal structural defects preventing cell separation at this stage of division. This is very different than the phenotype observed in fla1 mutants, where disruption of flagellum attachment structures prevents the initiation of cytokinesis (30). Flagellar motility can influence positioning of the kinetoplast (unpublished observation), which is connected to the flagellar basal body (39), and it is possible that indirect effects also contribute to the cytokinesis defect, particularly several days postinduction. However, the finding that physical forces provided by mechanical rotation of cultures completely rescues the cytokinesis defect demonstrates that no physical barrier precludes cell separation and supports the idea that physical forces provided by the flagellar beat contribute to cell division. Mechanical forces also contribute to cytokinesis in animal cells (50). In wild-type trypanosomes at the final stage of cytokinesis, daughter cells are opposing one another, with their flagella pointing in opposite directions (video S13 in the supplemental material) (58). Since there is no compelling evidence for an actin/myosin II contractile ring at the cleavage furrow in T. brucei (15), exploiting directional and rotational pulling forces imparted by flagellar beating would be a convenient way to aid in the final separation of dividing cells. A similar phenomenon, termed rotokinesis, whereby ciliary driven cell motility and perhaps cell rotation assists in the separation of daughter cells has been described in Tetrahymena thermophila (8). Note that a universal feature of trypanosome motility mutants that fail in cytokinesis is that they do not rotate. In contrast, flagellar beat drives vigorous cellular rotation in wild-type cells and in trypanin single mutants, which divide normally (Table 2 and compare videos S13 and S14 in the supplemental material) (26).

Defects in ciliary motility are linked to wide variety of inherited human diseases (2, 27, 63) and a better understanding of flagellar motility is critical for the development of new strategies to treat these diseases. Although the structural features of the flagellum have been studied extensively and are considered to be well conserved, regulatory mechanisms that control flagellar beat are not well understood in any organism. In the present study, we demonstrate that the DRC is an evolutionarily conserved dynein regulatory system and that altered orientation of central pair microtubules is linked to cell motility defects in vivo. We also provide the first evidence that flagellar motility is required for cell division in T. brucei, the causative agent of African sleeping sickness. This suggests that the numerous enzymatic activities that drive the beating of the eukaryotic flagellum, such as ATPases, phosphatases, and kinases (48), represent candidate drug targets for treatment of this fatal disease.