Cilia-organizing basal bodies (BBs) are microtubule scaffolds that are visibly asymmetrical because they have attached auxiliary structures, such as striated fibers. In multiciliated cells, BB orientation aligns to ensure coherent ciliary beating, but the mechanisms that maintain BB orientation are unclear. For the first time in Tetrahymena thermophila, we use comparative whole-genome sequencing to identify the mutation in the BB disorientation mutant disA-1. disA-1 abolishes the localization of the novel protein DisAp to T. thermophila striated fibers (kinetodesmal fibers; KFs), which is consistent with DisAp’s similarity to the striated fiber protein SF-assemblin. We demonstrate that DisAp is required for KFs to elongate and to resist BB disorientation in response to ciliary forces. Newly formed BBs move along KFs as they approach their cortical attachment sites. However, because they contain short KFs that are rotated, BBs in disA-1 cells display aberrant spacing and disorientation. Therefore, DisAp is a novel KF component that is essential for force-dependent KF elongation and BB orientation in multiciliary arrays.
Motile cilia are whiplike projections that generate hydrodynamic force. Cilia-generated fluid flow is required for symmetry breaking during embryogenesis, mucus clearance, cerebrospinal fluid flow, and the directed movement of unicellular organisms (Marshall and Kintner, 2008). One cycle of ciliary beating constitutes a power stroke and a subsequent recovery stroke. Thus, ciliary beating is directional, and to produce coherent fluid flow, multiple cilia must orient their beating along a common plane, which is typically the cell’s anterior–posterior axis. The importance of proper cilia orientation is underscored by the observation that cilia orientation defects accompany primary cilia dyskinesias, a devastating class of genetic disorders (Rayner et al., 1996).
Cilia are organized by cylindrical microtubule scaffolds called basal bodies (BBs) that dock at the cell cortex (Jana et al., 2014). BBs are innately asymmetric and their polarity is reflected in the attachment of auxiliary structures, including striated fibers (Allen, 1969; Pearson, 2014). Thus, BBs have a specific orientation that determines the direction of ciliary beating (Tamm et al., 1975; Gibbons, 1981; Hoops et al., 1984). BBs with improper orientation relative to the cellular anterior–posterior axis will disrupt cilia-generated fluid flow. The mechanisms that organize and maintain BB orientation remain ill-defined.
Striated fibers project asymmetrically from BBs and influence BB positioning by an unknown mechanism (Allen, 1967; Wright et al., 1983; Hoops et al., 1984). SF-assemblin and rootletin are coiled-coil proteins that self-organize into filamentous fiber structures and constitute major structural components of striated fibers in protists and vertebrates, respectively (Lechtreck and Melkonian, 1991; Yang et al., 2002), although other proteins are also present (Lechtreck and Melkonian, 1998; Park et al., 2008; Chien et al., 2013). Moreover, striated fibers display dynamic assembly and disassembly (Salisbury et al., 1984; Sperling et al., 1991; Francia et al., 2012). Thus, striated fibers are complex and dynamic structures of which our molecular understanding is limited.
Unicellular ciliates, such as Tetrahymena thermophila, and multiciliated vertebrate cells harbor hundreds of cilia organized into ciliary arrays. Ciliary array BBs exhibit evolutionarily conserved striated fiber placement directly opposite the cilium’s power stroke (Allen, 1969; Peraldi-Roux et al., 1991; Frankel, 1999). The Tetrahymena striated fiber, the kinetodesmal fiber (KF), emanates close to the BB’s base and terminates within or directly underneath the membrane-skeletal layer near the adjacent anterior BB (Allen, 1967, 1969). The apposition of the KF and the postciliary microtubules from the anterior BB supports speculation that KFs stabilize ciliary rows by providing a physical linkage between neighboring ciliary units and by linkage to subcortical structures (Allen, 1967; Iftode and Fleury-Aubusson, 2003; Wloga and Frankel, 2012). Although this hypothesis has been strengthened by observations in Chlamydomonas reinhardtii (Wright et al., 1983; Hoops et al., 1984), a mechanistic understanding of how striated fibers organize ciliary arrays and respond to and resist mechanical forces has not been established.
Results and discussion
DisAp localizes to KFs and orients BBs
disA-1 is a single-locus, recessive mutation generated in a mutagenesis screen for T. thermophila BB organization defects (Frankel, 1979, 2008; Jerka-Dziadosz et al., 1995). DISA organizes BBs into ciliary rows, but is dispensable for global cellular polarity and for ciliogenesis (Fig. 1 A; Jerka-Dziadosz et al., 1995). The disA-1 gene was identified using comparative genome sequence analysis with next-generation sequencing (Fig. S1, A and C). This approach identified a splice acceptor site mutation in intron 1 of a novel gene (TTHERM_00941400), which results in a severely truncated protein (Fig. 1 B and Fig. S1 B). The gene encodes a protein (DisAp) containing a similarity to the SF-assemblin consensus domain (Fig. 1 C; Lechtreck and Melkonian, 1998). Although the faint resemblance to SF-assemblin alerted us to a potential role in KF structure (pfam06705; BLASTp query of the Conserved Domain Database), phylogenetic analysis revealed that DisAp is a member of a distinct family of proteins conserved among ciliates, with seven paralogues in T. thermophila. Although this family includes other proteins with proposed roles in BB function (i.e., Bbc29p and Bbc39p; Kilburn et al., 2007), it does not include SF-assemblin. Therefore any shared function between DisAp and SF-assemblin may reflect convergent evolution, similar to previous observations for dynamin-like proteins in ciliates and metazoans (Elde et al., 2005). Introduction of TTHERM_00941400 into disA-1 cells rescues BB disorganization (Fig. 1 D). Thus, BB disorganization in disA-1 is caused by the mutation of DisAp. Moreover, our discovery of DisAp provides proof-of-principle for the combined use of Tetrahymena forward genetic screens and next-generation sequencing to identify novel BB mutants.
SF-assemblin is the major component of algal striated fibers (Lechtreck and Melkonian, 1991). Striated fibers in Paramecium tetraurelia are composed of multiple polypeptides (Sperling, 1989), suggesting that the family of SF-assemblin genes expanded in ciliates. Indeed, DISA belongs to a family of genes that has undergone frequent duplications, especially evident in P. tetraurelia, and appears to be conserved in Giardia, a lineage distantly related to ciliates (Fig. S1 D and Table S1). However, SF-assemblin from C. reinhardtii does not associate with the clade, as assessed by reciprocal BLAST searches that failed to recover identity between these proteins (unpublished data). Therefore, to test whether DisAp localizes to T. thermophila KFs, wild-type (WT) GFP-DisAp and mutant disA-1 GFP-disA-1p were localized relative to BBs and KFs in otherwise WT cells. GFP-DisAp localizes to punctae along ciliary rows that colocalize with the proximal portion of the KF (Fig. 1, E and F). In contrast, mutant GFP-disA-1p does not localize to KFs (Fig. 1 E). GFP-DisAp is restricted to ciliary row BBs and is absent from oral apparatus BBs (Fig. 1 E, arrows), which form normally in disA-1 cells (Fig. 1 D, arrows).
DisAp’s localization near the base of the KF suggested that it localizes to a discrete domain within the KF. Endogenously tagged DisAp-mCherry also localizes to the proximal portion of the KF (Fig. 1 F). Signal for DisAp is anterior to BBs and decreased below 50% ∼500 nm before KF intensity declined to 50%. Consistently, DisAp localized by immuno-EM clustered near the base of the KF (Fig. 1 G, yellow arrows; and Fig. S1 E). We also detected DisAp adjacent to the KF (Fig. 1 G, white arrows), which may reflect a population that has not incorporated into the KF. Thus, DisAp localizes to a domain at the proximal portion of the KF and is phylogenetically distinct from SF-assemblin.
DisAp loss disrupts BB orientation and prevents temperature-induced KF elongation
In disA-1 cells, KFs are disoriented relative to the cellular anterior–posterior axis. Therefore, DisAp could either specify the location of KF attachment to BBs or prevent BB rotation. Normally, KFs from adjacent BBs are aligned along a common axis and are oriented ∼180° from postciliary microtubules (Fig. 2 A; Allen, 1969). This KF placement is analogous to vertebrate BBs where the basal foot microtubules are positioned ∼180° from the striated rootlet (Steinman, 1968; Peraldi-Roux et al., 1991). In disA-1 cells, KFs are positioned ∼180° from the postciliary microtubules (Fig. 2 A), which suggests that DisAp prevents BB rotation and does not affect accessory structure placement. Moreover, disA-1 KFs are shorter than WT KFs (Fig. 2 A). Thus, DisAp prevents BB rotation and is required to establish and/or maintain appropriate KF length. We propose that DisAp functions as a regulator of KF elongation.
BB orientation defects in disA-1 cells are exacerbated by elevated temperature (Fig. 2 B; Jerka-Dziadosz et al., 1995). Thus, short KFs might allow temperature-induced BB rotation. If true, long KFs should prevent BB rotation. One prediction from this inference is that KFs elongate at elevated temperatures to resist BB rotation. To test this, we developed a semiautomated image analysis routine to measure KF length as well as BB orientation, and we assessed these parameters after shifting G1-arrested cells to 37°C and releasing them into the cell cycle (Fig. 2 C). Before the temperature shift, WT cells had a mean KF length of 1.10 µm (Fig. 2 D). After the temperature shift, KF elongation reached 1.38 µm in length after 24 h (Fig. 2 D and Fig. S2 A). disA-1 KFs were approximately half as long (0.49 µm), and elongation was more gradual than in WT cells (Fig. 2 D and Fig. S2 B). Unlike WT, disA-1 cells displayed a time-dependent randomization of BB orientation (Fig. 2, E and F; and Fig. S2, C and D). These experiments were performed after a starvation-induced G1 arrest. Because starvation affects cortical organization (Nelsen and Debault, 1978), this could complicate our analyses. However, increased temperature in cycling cells that were not synchronized in G1 also caused WT KF elongation (Fig. 3, A and B) and increased disA-1 BB disorientation (Fig. 3 C). Thus, increased temperature, and not starvation, promotes DisAp-dependent KF elongation and increases the severity of BB disorientation in disA-1 cells. Collectively, these data uncover a novel relationship between KF length and BB orientation. First, the KF is dynamic and elongates in response to elevated temperature. Second, normal KF length requires DisAp. When the KF length is impaired, BBs are susceptible to rotation. We next investigated how temperature induces these changes in BB morphology.
DisAp confers resistance to mechanical forces produced by ciliary beating
Elevated temperature increases cilia beat frequency and cell swimming speed (Goto et al., 1982; Pearson et al., 2009), which confers greater cilia-generated forces on BBs (Bayless et al., 2012). We explored whether temperature-induced increases in force in disA-1 corresponds with the observed BB disorientation by quantifying cellular swim speeds at differing temperatures (Fig. 3, D and E). At 25°C, WT cells swam at 272 µm/s; this increased to 392 µm/s after a 10-min incubation at 37°C (acute) but decreased to an intermediate level (315 µm/s) after prolonged 24-h incubation at 37°C (chronic). disA-1 cells at 25°C exhibited a reduced swimming rate relative to WT cells (123 µm/s). Acute temperature shift increased the velocity (228 µm/s). However, unlike WT cells, increased motility was not sustained, as chronic maintenance at 37°C decreased the swimming rate below that of disA-1 cells grown at 25°C. This motility defect parallels disA-1 BB disorganization, with prolonged growth at 37°C causing more severe BB disorientation. The initial increase in swim speed in disA-1 cells shifted to 37°C for 10 min (acute) is likely the result of increased beat frequency. However, prolonged exposure to increased beating forces may drive BB disorientation, thereby decreasing the effective rate of cell swimming.
Cilia-generated force increases WT KF length and disA-1 BB disorientation
We next tested whether increases in ciliary force influence KF elongation and BB orientation independent of temperature changes. The drag forces (physical resistance) that cilia experience can be increased by increasing their environmental viscosity with polymers (Spoon et al., 1977; Jung et al., 2014). In cycling cells cultured in high viscosity media (polyethylene oxide [PEO]) at 25°C, WT KFs elongated (Fig. 3 F) and disA-1 cells increased BB disorientation (Fig. 3 G). Moreover, high viscosity media also caused G1-arrested WT cells to undergo KF elongation (Fig. 3 F) and disA-1 cells to exhibit randomization of BB orientation (Fig. 3 G). Because G1-arrested cells do not assemble new BBs, BB assembly is not required for KF elongation or BB disorientation. Finally, increasing ciliary beat frequency with the cAMP agonist IBMX (Hennessey and Lampert, 2012) lengthened WT KFs and increased disA-1 BB orientation defects (Fig. S3, B and C), and when high temperature shift was accentuated with increased viscosity, an additive effect was observed (Fig. S3, D–F). Thus, increased ciliary-generated force triggers KF elongation. In the absence of KF elongation, as observed for disA-1, enhanced ciliary forces disrupts BB orientation.
Because cilia-generated force leads to KF elongation, we asked whether a reduction in ciliary beating prevents temperature-induced KF elongation. In the presence of NiCl2 or vanadate, WT temperature-induced KF elongation at 2 and 8 h was abolished (Fig. 4 A and Fig. S3 G), which suggests that KFs elongate due to cilia-generated forces. Consistent with this, growth at 15°C slows cell swimming (Beveridge et al., 2010) and reduces KF length (Fig. S3 G). Moreover, the disorientation observed upon shifting disA-1 cells to 37°C was rescued by reducing ciliary beating with either NiCl2 or vanadate (Fig. 4, B and C). Similarly, growth at 15°C slightly reduced disA-1 BB disorientation (Fig. S3 H; not statistically significant). Thus, cilia-generated force is both necessary and sufficient to increase BB disorientation in disA-1 cells, and BB rotation is resisted by DisAp-mediated KF elongation.
KFs guide nascent BBs and maintain the position of mature BBs
KFs terminate subjacent to the cortical membranes in the membrane-skeletal layer, where stable attachment of the KF may stabilize BBs against rotation. To determine whether KF elongation increases contacts between the KFs and the membrane-skeletal layer, we used three-dimensional image averaging to determine the mean length of the KF that lies near to the cell cortex. This position is defined by centrin, which marks the distal end of the BB near the membrane-skeletal layer (Stemm-Wolf et al., 2005). In WT cells at 25°C, the KF full-width at half maximum (FWHM) intensity above the plane of centrin was 1.09 µm long, which increased to 1.28 µm upon shifting to 37°C for 24 h (Fig. 5 A). In disA-1 cells cultured at 25°C, the KF FWHM was 0.57 µm, and it decreased to 0.51 µm after shifting to 37°C (Fig. 5 B). These results argue that force-dependent KF elongation augments the contact between the KF and anchoring structures in the cell cortex.
BB orientation in ciliates is propagated via a nongenetic process termed cytotaxis, which relies upon preexisting structures, such as old BBs, to constrain the position and orientation of newly arising structures, such as new BBs (Sonneborn, 1964; Beisson and Sonneborn, 1965; Beisson, 2008). Interactions between BBs and KFs or striated rootlets are proposed to organize the even spacing of BBs (Allen, 1969; Wright et al., 1983; Hoops et al., 1984; Lechtreck et al., 2002; Iftode and Fleury-Aubusson, 2003). In ciliates, these interactions occur between the KF and the postciliary microtubules of adjacent BBs in a ciliary row (Fig. 5 C). Nascent BBs are assembled at a mother BB and then transported along the mother BB’s KF to separate the daughter from the mother (Fig. 5 C). In disA-1, the association between neighboring BBs and the KF is generally preserved (Fig. 5 D), which suggests that DisAp is not essential to link adjacent BBs to the KF. However, because disA-1 KFs are short and disoriented with respect to the cellular anterior–posterior axis, BB separation along the KF leads to clusters of closely spaced BBs oriented along a shared axis, each of which deviates from the cellular polarity (Fig. 5 D). Thus, similar to striated fiber-dependent centrosome cohesion and daughter cell positioning (Bahe et al., 2005; Francia et al., 2012), KFs actively position BBs in multiciliary arrays. Moreover, by ensuring that KFs reach an appropriate length, which prevents BB rotation, DisAp allows cytotaxis to perpetuate accurate cortical patterning.
We show that the KF is a major component of the structural environment into which nascent BBs are born. In addition, a genetic input, DISA, is required to maintain this environment. We have identified situations (DisAp-deficient) in which cilia-generated forces weaken and partially abolish cytotaxis. This expands upon the concept of structural inheritance to demonstrate that it is both plastic and subservient to the forces that act on BBs.
We demonstrate that the length of the striated fiber in T. thermophila, the KF, is responsive to forces generated by cilia. Furthermore, KF elongation stabilizes BB orientation, ensuring ciliary alignment and coherent fluid flow. Through next-generation sequencing, we identified DISA as a gene responsible for BB organization whose protein is required for KF elongation. Finally, the stability of BB orientation and KF length are important for the propagation of the structural order in cells.
How the forces generated by cilia are sensed and then translated into KF length regulation remains to be determined, and the site of force detection, whether it be the BB or the KF, is also unknown. Because DisAp localizes near the BB and is important for KF elongation, it is an attractive target for force response. Furthermore, our results extend beyond cortical patterning in ciliates. In vertebrates, the striated rootlet, which is analogous to the KF, plays a prominent role in stabilizing the orientation of the ciliary unit (Chien et al., 2013). Therefore, our study raises the intriguing possibility that force sensing and response by BB-associated striated fibers is a conserved mechanism that has independently evolved in different eukaryotic lineages to couple ciliary forces to BB orientation.
Materials and methods
T. thermophila cells were grown in 2% SPP media (2% proteose peptone, 0.2% glucose, 0.1% yeast extract, and 0.003% Fe-EDTA) at the indicated temperatures (either 15°, 25°C, or 37°C). For all cycling cell studies, cells were analyzed at mid-log phase (density between 105 and 4 × 105 cells/ml) as determined using a Coulter Counter Z1 (Beckman Coulter). For starvation experiments, cells were arrested in the G1 phase of the cell cycle by washing and culturing in 10 mM Tris-HCl, pH 7.4, for 18–24 h. For microscopy experiments, analyses were restricted to nondividing cells as judged by those lacking an oral primordium. To expose cells to increased viscosity, Tetrahymena were propagated in 2× SPP supplemented with an equal volume of 7.5% polyethylene oxide (mol wt 900,000; Acros Organics), which was prepared in ddH2O by gentle mixing at 37°C for 24–48 h. Alternatively, 2% SPP was supplemented with 750 nm of the phosphodiesterase inhibitor 3-isobutyl-1-methylxanthine (IBMX; Sigma-Aldrich), which increases cAMP levels and cilia beat frequency. To expose cultures to decreased forces, Tetrahymena were propagated in 2% SPP supplemented with 250 µm NiCl2 (Sigma-Aldrich), which was added directly to the culture vessel from a 5-M stock. Dynein-dependent ciliary beating is inhibited by NiCl2, which blocks plasma membrane calcium channels and directly inhibits dynein motors (Larsen and Satir, 1991). Alternatively, 2% SPP was supplemented with 750 µm sodium orthovanadate (Sigma-Aldrich; Gibbons et al., 1978; Nilsson, 1999), which was added directly to the culture vessel from a 100-mm stock (pH 10).
Plasmids and Tetrahymena strain construction
To rescue the disA-1 phenotype with WT DISA, disA-1 cells (IA217) were transformed with the DISA ORF, and flanking sequences were inserted exogenously into the genome at RPL29. Specifically, DISA was PCR amplified from WT (B2086) cells using oligos (5′-CGCTGCAGAAAGATAGATGCTTGCTTGC-3′ and 5′-CGGAGCTCGCTGTATTCTAAAGTTCAAG-3′) and cloned into pBSMCSCHX. After release with Blp1 digest, the rescue construct was biolistically transformed into disA-1 cells and selected for cycloheximide resistance and rescue of the disA-1 phenotype.
GFP-DISA and GFP-disA-1 were cloned into pNEO2-MTTpr-GFP that inserts an MTT-GFP cassette upstream of the endogenous gene (Winey et al., 2012). The cassette containing either WT or disA-1 mutant sequence was inserted into otherwise WT cells. 0.6 kb of sequence upstream of DISA was PCR amplified (5′-CGGAGCTCCTGTTTAAAATTAAGCATGCTC-3′ and 5′-CGGGATTCGCCTTATTAACCGTTTCCTT-3′) and cloned into pNEO2-MTTpr-GFP. Next, a 0.6-kb sequence of either DISA or disA-1 was PCR amplified (5′-CGCTCGAGATGTCTGCTTTCGGCTCTCC-3′ and 5′-CGGGGCCCCTTAATTTCTTTACCCTTTC-3′) and cloned into the plasmid to produce either pNEO2-MTT-GFP-DISA or pNEO2-MTT-GFP-disA-1.
A DisAp-mCherry strain was constructed. The p4T2-1:DisA:mCherry cassette integrates at the endogenous DISA locus and remains under the control of the endogenous promoter. p4T2-1:DISA:mCherry was generated by PCR amplifying (5′-CGGGATCCAAAATTGCAACTATCTAAAC-3′ and 5′-CGGAGCTCTCAAGTGAGCTTTAACTATC-3′) and cloning the final 0.6 kb of DISA without the TGA stop codon into p4T2-1-mCherryLAP (Winey et al., 2012). A 0.9-kb fragment downstream of the TGA stop codon (5′-CGGGTACCAGAAAATCATATTGAAACAC-3′ and 5′-CGGAATTCTTGACGGAAGTTCTTATCAATTCTCTAGCAAGTG-3′) was then cloned into the plasmid to create p4T2-1:DISA:mCherry. This plasmid contains NEO2 drug selection.
Next-generation sequencing and identification of disA-1
To generate a backcross of the disA-1 mutation, IA217 (disA-1) and B1868 (WT) Tetrahymena lines were crossed to produce micronuclear heterozygous F1 progeny. Two F1 clones of different mating types (F1.1 and F1.8) were then mated to produce 18 F2 disA-1 mutant lines. Total genomic DNA was purified from each line using a urea-SDS lysis and phenol:chloroform extraction (Gaertig et al., 1994) and measured using a Qubit Fluorometer (Invitrogen). Equal DNA was pooled for all 18 F2 clones to produce a single Illumina Tru-seq library. A second DNA preparation of B1868 was also prepared. Using Illumina’s standard TruSeq DNA library preparation, DNA was first sheared to a 300–400-bp distribution using sonication, and then end-repaired and A-tailed using a combination of T4 polynucleotide kinase (PNK), T4 DNA polymerase, and Klenow. A-tailed DNA was then ligated to Illumina Truseq adapters, and further independently indexed using PCR. Both libraries were size selected to remove adapter and PCR dimers, pooled, and co-sequenced on a single lane of the Illumina HiSeq 2000 using a 2 × 100 base pair format. Postsequencing bioinformatics were used to separate sequences from each of the two libraries using the unique indexes incorporated during the library process.
The pipeline for mapping and variant calling in the data are reported in a publicly available Wiki (https://github.com/jewmanchue/pooled-mapping). To improve read quality, 10 and 5 bp were trimmed from the 5′ and 3′ ends, respectively. Reads were locally aligned simultaneously to two reference sequences (micronucleus: tetrahymena_thermophila_sb210__mic__2_supercontigs; and macronucleus: JCVI-TTA1-2.2) using bowtie-2.0.0. (Langmead and Salzberg, 2012). Discordant mate pairs were discarded. The reads mapped to the genome were >98% for B1868 control and >95% for the disA-1 F2. Sequencing coverage for the entire genome was 0.96 and 0.60 for the B1868 and disA-1 strains, respectively. Alignment cleanup was performed using Samtools to remove PCR duplicates GenomeAnalysisTK-2.4-7 was used to realign indels and polish alignments (Li et al., 2009; McKenna et al., 2010; DePristo et al., 2011). Joint variant calling for the disA-1 F2 and the B1868 control pool was performed using SNVerPool (Wei et al., 2011).
Heuristic filters were applied to the 8,795,723 mutations (SNVs, 7,346,955; insertions, 648,913; deletions, 799,855) using the Genotype-Phenotype Association Toolkit (https://github.com/jewmanchue/vcflib/wiki). Positions in the genome with a depth <5 in the B1868 or disA-1 mutant pools were removed. The second filter removed positions where the B1868 pool contained the nonreference allele. The last filter leniently enforced a recessive model by requiring the frequency of the nonreference allele in the disA-1 mutant pool to be >0.75. A filter of 0.75 was used to avoid false negatives created from sequencing errors. However, the disA-1 mutation did have an allele frequency of 1.0. After filtering, only 206 mutations were remaining, 26 of which were homozygous nonreference (allele frequency of one) in the disA-1 mutant pool (Fig. S1 C). We focused on the nine mutations mapped to the macronuclear sequence. Nine candidate positions for the disA-1 mutation were identified and narrowed down by searching for proteins containing domains commonly associated with BBs and their auxiliary structures. Hand annotation revealed that one of the mutations was a splice site mutation in the 3′ acceptor site (G to A) in TTHERM_00941400. This mutation was found in both the macronuclear and micronuclear sequences. The mutation was also confirmed by both PCR of the genomic region and cDNA of TTHERM_00941400 and sequencing.
The amino acid sequence of DisAp from T. thermophila (XP_001026900) was used in a protein–protein blast query (http://www.ncbi.nlm.nih.gov) to identify related sequences (Table S1). Alignments were generated with ClustalW2 (http://www.ebi.ac.uk) and trimmed by eye to eliminate insertions and deletions. Model fitting and tree inference of the alignment was performed with Mr. Bayes v3.2.2 (Ronquist and Huelsenbeck, 2003). The “rtrev” amino acid substitution model was best supported by the data and used for inferring the tree. After the burn-in phase, the remainder of 500,000 generations of Markov chain Monte Carlo analysis were considered for inference of the tree. The 50% majority rule consensus tree was generated and visualized with FigTree v1.4.
For immuno-cytochemical analyses, 1–3 × 105 cells were pelleted at 1,500 g in a 1.5-ml Eppendorf tube and fixed for 20–30 min with 1.5 ml of 70% ethanol + 0.2% Triton X-100. Cells were washed with 10 mM Tris-buffered saline and blocked overnight at 4°C in 1% BSA in 10 mM TBS. Cells were immunostained by incubating overnight at 4°C in primary antibody (mouse anti-KF [5D8], 1:400; Jerka-Dziadosz et al., 1995; rabbit anti-centrin, 1:2,000, a gift from A. Stemm-Wolf and M. Winey, University of Colorado Boulder, Boulder, CO; Stemm-Wolf et al., 2005; rabbit anti–α tubulin, DM1a; Sigma-Aldrich) followed by a 1-h incubation at room temperature in secondary antibody (goat anti–mouse Alexa Fluor 594, 1:2,000; goat anti–rabbit Alexa Fluor 488, 1:2,000; goat anti–Alexa Fluor 647, 1:2,000; Invitrogen). Cells were mounted in Citifluor mounting media (Citifluor LTD) using #1.5 coverslips and sealed with nail polish. All antibodies were diluted in 1% BSA/TBS. Cells were washed (3 × 5 min) with 1% BSA/TBS after primary and secondary antibody incubations.
For the localization experiments in Fig. 1, an inverted microscope (Ti Eclipse; Nikon) with a 100× Plan-Apochromat (NA 1.4) objective lens (Nikon) was used. Images were captured with an electron-multiplying charge-coupled device (EMCCD) 888E camera (iXon; Andor Technology). For all other experiments, confocal microscopy was performed using an inverted microscope (Ti Eclipse) with a 100× Plan-Apochromat (NA 1.43) objective lens (Nikon) and a Swept Field confocal scan head (Prairie Technologies). Confocal images were acquired in slit mode with a slit size of 35 µm and a z-step size of 200 nm, and detected with a charge-coupled device (CCD) camera (Clara; Andor Technology). Images were acquired with Elements software (Nikon) and all fixed cells were imaged at room temperature.
Transmission EM (TEM)
EM was performed as described previously (Pearson et al., 2009; Bayless et al., 2012). A Tetrahymena strain expressing endogenous C-terminal DisAp-mCherry was grown to mid-log phase and then prepared for immuno-EM using high-pressure freezing and freeze substitution (HPF-FS; Dahl and Staehelin, 1989; Meehl et al., 2009). T. thermophila cells were pelleted, high-pressure frozen (HPM-010; Bal-Tec), freeze substituted in 0.25% glutaraldehyde/0.1% uranyl acetate in acetone, and embedded in Lowicryl HM20. 60-nm serial sections were cut and put on nickel slot grids, blocked with 1% milk in PBS–Tween 20, and incubated with anti-mCherry (rabbit polyclonal; a gift from I. Cheeseman, Massachusetts Institute of Technology, Cambridge, MA) at 1:100. 15 nm of gold-conjugated secondary antibody was applied to the grids at a dilution of 1:20 (Ted Pella). Grids were poststained with 2% uranyl acetate and lead citrate.TtBld10 was then localized in 60-nm sections using TEM. Images were collected using a Philips CM10 electron microscope (Philips) equipped with a Gatan BioScan2 CCD camera (Gatan). For structural analyses of disA-1 BB defects, disA-1 and control (B1868) cells were subjected to HPF-FS after growth at 25°C.
Tetrahymena motility measurements
Free-swimming Tetrahymena in glass-bottom dishes were imaged using a 20× objective lens (pixel size, 330 nm) and transmitted light on the confocal microscope (see “Light microscopy”). For each field of view, images were captured at ∼170-ms intervals for a total of 30 s. To track motility paths, we marked the anterior tip of individual Tetrahymena that displayed directional motility for the duration of their swim path while they remained in focus. Care was taken to avoid Tetrahymena at the glass surface. Tracking analysis was facilitated by the MTrackJ (Meijering et al., 2012) plugin bundled with FIJI (Schindelin et al., 2012).
Image analysis: KF length and BB orientation
KF length and BB orientation quantification were performed in a semiautomated fashion using the macro scripting language and plugins contained within the FIJI build of ImageJ. Image stacks were preprocessed with a Laplacian of Gaussian filter (LOG; radius, 1 pixel) to reduce noise and enhance feature edges. 32-bit LOG stacks were inverted, their contrast was adjusted (minimum = mode pixel intensity + (1/2) × standard deviation pixel intensity; maximum = maximum pixel intensity), and the images were merged to create an 8-bit RGB image stack. To quantify KF length for individual cells, 10 KFs with clear separation from neighboring KFs were manually measured by tracing with the freehand line tool. To quantify BB orientation, a box (10 µm wide × 5 µm tall) was placed in the center of a Tetrahymena cell and angular measurements were made for 10 BBs within that box. For each BB, the angular measurement represents the angle between the tip of the KF and the anterior pole of the cell (Fig. 2 C). For each cell, the mean vector and the length of the mean vector (R value) for the 10 measured BBs were calculated using circular statistics and displayed on polar plots. Each cell was measured twice, once on each side; thus each cell produced two R values. On the polar plots the mean vector is represented by the direction of the arrow. Variance in the mean vector is represented by the length of the arrow. An arrow length of 1 indicates no variation in the mean vector and an arrow length of 0 represents pure randomness in the mean vector. On the polar plots, the dashed circles represent 0.2 arbitrary units of R value. To compare the amount of BB orientation defects across different populations of cells, the mean R value for the cell population was determined in linear space. Circular statistics were calculated using the ORIANA circular statistics suite (Kovach Computing Services).
Image analysis: fluorescent image averaging
The brightest centrin (BB) voxel for an individual BB was determined. A 5-µm box was centered over this voxel in the x, y, and z dimensions. The raw BB and KF image stacks were cropped in the xy dimension using the 5-µm box, and they were cropped in the z dimension by taking five slices below the slice containing the brightest BB voxel and five slices above the brightest BB voxel (11 slices total; 3.3 µm). Next, cropped stacks were rotated so that the tip of the KF was aligned with a straight line that ran down the middle of the 5-µm box and passed through the brightest centrin pixel. This procedure was performed on 100 BBs from 10 different cells for each condition. The raw images used for averaging were part of the 0 h and 24 h time points (SPP condition) of the dataset used in Fig. S3 (D and E). To create the average image stack, individual image stacks were averaged on a per-slice basis. All image cropping was performed with FIJI using the crop, rotate, and duplicate stack commands. The yz images were created by rotating the averaged image stacks in three dimensions using the TransformJ plugin.
All linear statistical analyses were performed in Excel (Microsoft). All tests for significance were unpaired, two-tailed t tests. All error bars indicate SEM. Statistical significance was set at P < 0.01.
Online supplemental material
Fig. S1 shows the scheme used for the identification of the disA-1 mutation, a phylogenetic tree of related DisAp proteins, and immuno-EM confirming DisAp’s localization at the KF. Fig. S2 shows the frequency distributions of WT and disA-1 KF length and BB orientation upon temperature shift, which documents population-wide shifts in KF length and BB orientation. Fig. S3 shows that WT BBs are resistant to force-induced orientation defects, whereas additional force perturbations impact WT KF length and disA-1 BB orientation. Table S1 lists the Tetrahymena DISA-1 clade members.
We thank Alex Stemm-Wolf and Dr. Brian Mitchell for comments on the manuscript. We thank Dr. Ken Jones, Joe Brown, Dr. Bifeng Gao, and Katrina Diener for providing sequencing and preliminary analyses.
The sequencing was supported by a pilot grant from the University of Colorado Denver Genomics and Microarray Core made possible by the Genomics and Biostatistics/Bioinformatics Shared Resources of Colorado’s National Institutes of Health (NIH)/National Cancer Institute Cancer Center Support Grant P30CA046934. C.G. Pearson is supported by NIH-NIGMS GM0099820 and the Boettcher Webb-Waring Foundation. C.G. Pearson and N.C. Elde are supported by the Pew Biomedical Scholars Program.
The authors declare no competing financial interests.