Motor protein Kif6 regulates cilia motility and polarity in brain ependymal cells

Multiciliated cells (MCCs) have dozens of motile cilia that are planar polarized and coordinately beat to generate a directional flow across the epithelium (Brooks and Wallingford, 2014). Cilia-driven directional fluid flow maintains mucus clearance in the respiratory tract, transport of gametes in the reproductive system and circulation of cerebrospinal fluid (CSF) in the brain (Spassky and Meunier, 2017); however, recent studies have revealed subtle but important differences in these MCC populations (Konjikusic et al., 2018; Roberson et al., 2020).

Dyneins and kinesins are ATPase microtubule motor proteins that drive diverse motion in cilia and flagella (Gennerich and Vale, 2009). For example, axonemal dynein complexes, such as outer and inner dynein arms, are arrayed along microtubule doublets to generate a motive force for cilia beating (Ishikawa, 2017). However, while both kinesins and dyneins help facilitate intraflagellar transport of cargoes necessary to build and maintain cilia (Prevo et al., 2017), the function of several ciliary kinesins remains poorly defined (Konjikusic et al., 2021).

The kinesin-9 family, for example, comprises two subfamilies – kinesin-9A and kinesin-9B (Wickstead et al., 2010) – that have been implicated in cilia function but remain only cursorily studied. In Trypanosome brucei, both family members, KIF9A and KIF9B, are localized to the axoneme and their knockdown reduced flagellum motility (Demonchy et al., 2009). The KIF9A homologues KLP1 in Chlamydomonas reinhardtii and Kif9 in vertebrates have recently been shown to be active motor proteins that act in the central pair to facilitate ciliary beating (Konjikusic et al., 2023; Han et al., 2022).

Not much is known about members of the kinesin-9B subfamily. The kinesin-like protein Kif6 in vertebrates has been implicated in cardiovascular disease in humans (Ruiz-Ramos et al., 2015), although a meta-analysis of multiple case-control studies failed to find evidence of this association (Assimes et al., 2010). Instead, more-recent studies in fish, mouse and human did suggest a role for Kif6 in cilia in the brain (Gray et al., 2021; Konjikusic et al., 2018). In both zebrafish and mouse, loss of Kif6 function is associated with hydrocephalus and a decrease in the number of cilia within ventricles of the adult brain (Gray et al., 2021; Konjikusic et al., 2018); however, little else of Kif6 function regarding cell biology is known.

Here, we provide evidence that Kif6 functions as a canonical plus-end directed kinesin motor capable of slow, processive, microtubule-based movement in vitro. In mouse, loss of Kif6 function elicits a complex array of cilium-related phenotypes in the brain, including defects of cilia beating, CSF flow and planar polarization of basal bodies. Over time, ependymal cilia in Kif6 mutant mouse are lost, possibly due to defects in apical actin assembly and subsequent destabilization of basal bodies. Thus, our data suggest that Kif6 plays several roles in ependymal cell cilia homeostasis and neurodevelopment in mammals, and might shed light on Kif6-related human pathologies.

We previously reported a macrocephaly patient with a mutation in KIF6 that eliminates part of the coiled-coil stalk region and the C-terminal tail domain (Konjikusic et al., 2018), i.e. the putative cargo-binding domain (Piddini et al., 2001). We also generated mice with a similar mutation eliminating the C-terminal stalk and tail domains of Kif6, which lead to variably penetrant hydrocephalus and a reduction in the number of ependymal cell cilia (Konjikusic et al., 2018). Since the motor activity of Kif6 has never been determined, these findings led us to perform in vitro studies.

The alignment of the protein sequences of human (h), mouse (m), and Danio rerio (d) Kif6 proteins showed that the overall domain organization is highly conserved (Fig. 1A, Fig. S1A,B). When overexpressed in COS-7 cells, full-length (FL) mouse Kif6-EGFP (hereafter Kif6 FL-EGFP) did not localize to microtubules but, rather, showed a diffuse localization within the cytosol (Fig. S1C). In in vitro single-molecule motility assays, Kif6 FL-EGFP did not bind to taxol-stabilized microtubules (Fig. S1D), suggesting that full-length mKif6 is regulated by autoinhibition, which is common for kinesin proteins (Verhey and Hammond, 2009). To examine the motility properties of mKif6, we generated a truncated version of mKif6 (1-493) tagged to mNeonGreen [hereafter referred to as mKif6(1-493)-mNG], which includes the motor domain and the first two coiled-coil segments for homodimerization (Fig. 1A). When overexpressed in COS-7 cells, mKif6(1-493)-mNG was found both along microtubules and diffusely in the cytoplasm (Fig. 1B).

To visualize Kif6 movement directly, we carried out in vitro single-molecule motility assays and found that mKif6(1-493)-mNG moved processively along microtubules with a speed of 12.2±2 nm/s (mean±s.d.) (Fig. 1C–E, Movie 1). Similar speed (15.5±4.4 nm/s) was obtained with a truncated version of dKif6 (not shown). This slow rate of processive movement along microtubule tracks is similar to that reported previously for the related kinesin Kif9 (Konjikusic et al., 2023).

To test whether Kif6 protein moves processively towards the plus or the minus end of microtubules, we repeated the single-molecule motility assays in the presence of the kinesin-3 protein KIF1A, as a truncated version of this motor protein – KIF1A(1-393) that lacks amino acids (aa) 1–393 – is known to move with high speed and processivity towards the plus ends of microtubules (Hammond et al., 2009; Soppina et al., 2014). For these experiments, we used KIF1A(1-393)-Halo^JFX554, for which KIF1A(1-393) had been conjugated to the HaloTag and labeled with the HaloTag ligand JFX554. Given the differences in speed between the kinesin-3 KIF1A and the kinesin-6 proteins hKif6, mKif6 and dKif6, we first imaged motility of KIF1A along microtubules by using a rapid frame rate (ten frames per second) and then the motility of the KIF6 proteins along the same microtubule by using a slower frame rate (one frame every 3 s). Kymographs revealed that both mouse KIF6(1-493)-mNG (Fig. S2) and zebrafish KIF6(1-501)-mNG (not shown) moved in the same direction as KIF1A, demonstrating that they move processively towards the microtubule plus end.

We then expressed mKif6(1-493)-mNG in developing ependymal cells, where it colocalized to β-tubulin not only in cytoplasmic microtubules but also in cilia (Fig. 1F,I). While mKif6(1-493)-mNG accumulated at the tip of cilia (Fig. 1G,I), we were unable to observe any clear movement of mKif6(1-493)-mNG within these cilia (Fig. 1H, Movie 2). Together, these data suggest that, although the Kif6 motor domain derived from either zebrafish or mouse is able to bind to and move processively along microtubules, it cannot do so appreciably inside cilia; similar behavior has been shown for other KIF9 family kinesins, such as Chlamydomonas Klp1 and Xenopus Kif9 (Konjikusic et al., 2023; Han et al., 2022).

We next set out to define endogenous Kif6 protein localization using immunochemistry. Antibody specificity was confirmed using western blot analysis of primary ependymal cell lysates derived from WT and Kif6^{pG555fs/pG555fs} homozygous mutant (hereafter referred to as Kif6^pG555fs) mice. The Kif6^pG555fs mutation was predicted to truncate the Kif6 protein, thus removing the antigen region recognized by anti-Kif6 antibody. Western blotting confirmed a single band of Kif6 protein expressed in testis and developing short cilia and developed long cilia (Fig. 2A). Interestingly, we consistently observed multiple bands when using anti-Kif6 antibody in lysates derived from lateral ventricular (LV) wall tissue. Because LV wall tissue contains many cells other than pure populations of primary cultured ependymal cells, robust Kif6 expression might be masked in these lysates. Next, we enriched for ependyma from primary LV wall tissue and confirmed ependymal cell differentiation at day (D)0, D5, and D14. At D0, the precursor cells had a short Arl13b-positive cilium (primary cilia). At D5, the cultured dish included precursor cells and immature ependymal cells that had multiple short or elongating cilia. At D14, mature ependymal cells had multiple long cilia (Fig. 2B). We first observed Kif6 at D5 of differentiation, which is concomitant with the onset of MCC differentiation within cultured ependymal cells.

We next examined Kif6 localization in ependymal cells by immunostaining in vivo. At D5 we observed a Kif6 punctum close to a primary cilium, with increased accumulation at the base of both developing short cilia and developed long cilia (Fig. 2C). To resolve this accumulation of Kif6 in cilia we used super-resolution imaging of ependymal cell cultures at D14, showing puncta along the length of ciliary axonemes (Fig. 2D). We also observed Kif6 puncta at the apical cell cortex localized to the basal foot (BF) marker γ-tubulin (Fig. 2E) and partially colocalized to microtubules (β-tubulin) at apical cell cortex around the ciliary base, as reflected by the staining for γ-tubulin (Fig. 2F). The specificity of immunostaining with anti-Kif6 antibody was confirmed by the loss of signal at multicilia in D14 ependymal cells derived from Kif6^pG555fs mutant mice (Fig. 2G). Finally, exogenously expressed mKif6 tagged to mCherry at its N-terminal end (mCherry-Kif6) localized to multicilia and the apical cell cortex together with β-tubulin (Fig. 2H). These results suggest that endogenous Kif6 localizes to the apical tubulin cortex at the base of cilia and to within ciliary axonemes.

As noted, Kif6^pG555fs mice display variably penetrant hydrocephalus (Konjikusic et al., 2018) (Fig. 3A). We observed moderate hydrocephalus with enlarged LV wall tissue in 69.2% (nine out of 13) mice, while severe hydrocephalus, often associated with a disrupted LV wall tissue and the absence of ependymal cells, was observed in the remainder (Fig. 3B). Because ependymal cilia were present in the middle part of LV wall tissue in Kif6^pG555fs mice with moderate hydrocephalus (Fig. 3C,D; Fig. S3), we chose these for a more detailed analysis of ciliary phenotypes.

As ependymal cilia beating contributes to directional flow and CSF circulation in the mouse brain, we first observed the cilia-driven flow on the surface of LV wall tissue (Fig. 3E). Microbeads were placed on the surface of LV wall tissue at the middle part and flowed in a linear path along the anterior–dorsal region in wild-type (WT) LV wall tissue (Fig. 3E,F). However, bead displacement on Kif6^pG555fs ventricles was significantly slower than on those of WT (Fig. 3G) and more tortuous (Fig. 3F; Movies 3 and 4).

Because knockdown of Trypanosome Kif9B (a homolog of Kif6) in T. brucei causes motility defects in the flagellum (Demonchy et al., 2009), we next asked if the Kif6^pG555fs mutation also impairs ependymal cilia beating. To avoid potentially confounding effects of hydrocephalus and other environmental factors within brain ventricles, we evaluated the beating of cilia in primary cultured mouse ependymal cells. Live imaging of ependymal cells showed coordinated cilia beatings in WT and decreased cilia motility in Kif6^pG555fs mice (Fig. 3H, Movies 5-7). Consistent with the variable penetrance of hydrocephalus, we found that, at D14, Kif6^pG555fs ependymal cells had normal, faint or no motile cilia (Fig. 3I, Movie 6) and that the beating frequency of ependymal cilia, when present, was significantly decreased in Kif6^pG555fs compared with WT cilia (Fig. 3J). Interestingly, at D21, cilia beat frequency in Kif6^pG555fs mice was not significantly different compared with that in WT mice (Fig. 3J), suggesting that cilia beating in Kif6^pG555fs was restored as ependymal cells mature. Moreover, we found that the percentage of ependymal cells with immotile cilia was decreased in Kif6 mutant mice at D21 (Fig. S4A). The remaining Kif6^pG555fs ependymal cells at D21 had normal apical actin networks (Fig. S4B–D) and better cilium orientation (Fig. S4E,F) than the population of Kif6 mutant ependymal cells at D14. Thus, Kif6 appears to play a stage-dependent role in organizing directional ciliary beating.

As kinesins can form homo- and hetero-dimers, kinesins without the motor domain (motor-less) can act as dominant-negative proteins (Gelfand et al., 2001). To test this model, we transfected a motor-less Kif6 tagged to mCherry at its N-terminal end (hereafter referred to as mCherry-Kif6 ML) into ependymal cells. We observed that ependymal cells at D14 expressing mCherry-Kif6 ML significantly suppressed cilia beatings compared with cells expressing mCherry control (Fig. 3K,L, Movies 8 and 9). These results suggest that Kif6 regulates cilia motility in ependymal cells.

Cilia motility requires a radially arranged structure of nine doublet microtubules and axonemal dyneins in repeating patterns (Ma et al., 2019). To observe if ultrastructural defects may explain the decreased cilia beat frequency, we visualized cilia structures in WT and Kif6^pG555fs mutant mice by using transmission electron microscopy (TEM). TEM images showed a typical 9+2 arrangement – i.e. nine doublet microtubules surrounded by a central pair of singlet microtubules – with outer dynein arms in both WT and Kif6^pG555fs ependymal cilia at postnatal day (P)14 (Fig. 4A, Fig. S5). By using immunofluorescence we show that the outer dynein arm component Dnai1 was present in Kif6^pG555fs ependymal cell cilia as in WT (Fig. 4B,C). Decreased cilia length is associated with decreased cilia motility (Bottier et al., 2019); however, we found that the length of Kif6^pG555fs ependymal cilia, when present, was not significantly different from that of control cilia (Fig. 4D,E). Finding no overt structural defects in Kif6^pG555fs ependymal cilia, we turned our attention to the MCC apical cytoplasm.

Because Kif6 was localized to apical cytoplasmic microtubules, we next analyzed the apical microtubules that in ependymal cells form an organized network and align the basal bodies (BB's) (Werner et al., 2011; Kunimoto et al., 2012). Because Kif19A is known to depolymerize microtubules (Niwa et al., 2012), we first examined general microtubule organization in Kif6^pG555fs ependymal cells and found that overexpression of mCherry-Kif6 full length (mCherry-Kif6 FL) or dominant-negative mCherry-Kif6 ML had no effect on acetylated-tubulin, which correlates with stable microtubules in ependymal cell bodies (Fig. S6). Moreover, Kif6^pG555fs did not diminish the intensity of β-tubulin labeled microtubules accumulated around BBs (Fig. 5A,B).

Because the surface flow was disturbed on Kif6^pG555fs LV wall tissue (Fig. 3F), we next analyzed cilia polarity and orientation at single-cell (rotational polarity) and tissue level (translational polarity) (Mirzadeh et al., 2010). Rotational polarity was quantified by the orientation of the BF projection from BB. By using TEM for LV wall tissue at P14, we were able to show that the orientation of the BF-BB is perturbed in Kif6^pG555fs mice compared to the parallel orientation typically observed in WT mice (Fig. 5C). We also observed this rotational polarity defect in cultured ependymal cells derived from Kif6^pG555fs mouse mutants by immunostaining for the BB marker FGFR1 oncogene partner (FOP; officially known as CEP43) and the BF marker γ-tubulin (Fig. 5D). The orientation of BB-BF alignments in the plane of cultured ependymal cells at D14, was calculated as the mean resultant vector length (⁠⁠) and found to be significantly reduced in Kif6^pG555fs ependymal cells (Fig. 5E).

Ependymal cilia also display translational polarity, visible as the clustering of BB on LV wall tissue along the direction of CSF flow (Ohata and Alvarez-Buylla, 2016). We observed BB clustering within each ependymal cell at the anterior–dorsal area of LV wall tissue, finding them positioned ventrally (Fig. 5F). This translational polarity was disturbed in Kif6^pG555fs cells (Fig. 5G). BB clustering was quantified as vector angles from the center of the cell to the center of the cilium bundle base (Fig. 5H), and found to be more randomly distributed in Kif6^pG555fs LV wall tissue compared with the polarized distribution observed in WT tissue (Fig. 5I). BB clusters in Kif6^pG555fs mutants were close to the center of cell (Fig. 5J) and the directionality was significantly decreased (Fig. 5K).

It has been previously shown that tyrosinated tubulin, which indicates newly polymerized microtubule plus ends, is asymmetrically localized to apical junctions in ependymal cells, and that this microtubule polarization promotes cilia polarity (Takagishi et al., 2020). However, we found that Kif6^pG555fs LV wall tissue showed the normal asymmetric localization of tyrosinated tubulin (Fig. 5G). We also observed no difference in the intensity of staining for tyrosinated tubulin between Kif6^pG555fs and WT LV wall tissue (Fig. 5L). By contrast, the directionality of BB clusters towards polarized tyrosinated tubulin was significantly decreased in Kif6^pG555fs LV wall tissue (Fig. 5M,N). These results suggest that Kif6 affects cilia polarity independently of the apical microtubules.

Apical actin forms networks in MCCs, which aid in connecting BBs and contribute to their alignment, orientation and stabilization in MCCs (Werner et al., 2011; Mahuzier et al., 2018). By using phalloidin conjugated to a fluorescent dye to label actin filaments, we observed colocalization of accumulated apical actin with BBs in both WT and Kif6^pG555fs LV wall tissues (Fig. 6A). However, Kif6^pG555fs wall tissue showed a significant decrease in actin accumulation at with BB clusters – measured as phalloidin signal intensity at the cilia bundle base (Fig. 6B). Moreover, the area of BB clusters was significantly increased in Kif6^pG555fs ependymal cells (Fig. 6C) and dispersed across the apical cell cortex with decreased actin, compared to the tight accumulation of BBs in WT ependymal cells (Fig. 6A). Importantly, the loss of apical actin networks has been shown to induce BB dis-attachment in ependymal cells (Mahuzier et al., 2018). Accordingly, we found that the number of cilium bases, as counted by the number of BBs significantly decreased in Kif6^pG555fs ependymal cells (Fig. 6D). Because the BBs were not found in deeper cytoplasm (Fig. 6E), these results suggest that the defect of apical actin networks in Kif6^pG555fs cause destabilization of BBs and shedding of ependymal cilia.

Here, we described a complex phenotype of cilia-related defects in the ependymal cells of Kif6 mutant mice. We suggested that this complex phenotype emerges from pleiotropic defects in ependymal cells. Because we found that Kif6 actively moves along microtubules and is present in axonemes (Fig. 1), defects in cilia beating may result from the direct action of Kif6 in cilia motility. By contrast, because we demonstrated that Kif6 is required for apical actin assembly, its presence in the apical cytoplasm may explain defects in cilia polarity. Finally, we suggested that the defects in apical actin assembly also result in the loss of cilia from ependymal cells. Together, this constellation of ciliary defects results in defective CSF and moderate hydrocephalus in Kif6 mutant mice (Fig. 6F).

We have shown here that the kinesin-9B family member Kif6 is a slow processive motor in vitro, yet were unable to observe processive movement of the protein within cilia in vivo. Therefore, Kif6 closely resembles the related kinesin-9A protein Kif9 , as it undergoes slow processive motility in vitro but not inside cilia (Konjikusic et al., 2023). In addition, processive movement in axonemes of another kinesin-9A family member, i.e. the Chlamydomonas protein Klp1, also have not been observed (Han et al., 2022). One possibility is that the slow motility of kinesin-9 proteins evolved to permit nucleotide-dependent engagement and disengagement with central-pair microtubules and, thereby, contribute to ciliary bending in a manner analogous to how the non-processive axonemal dyneins engage with doublet microtubules (Han et al., 2022). Alternatively, the motility of kinesin-9 motor proteins in axonemes may be hindered, given that the central-pair microtubules are decorated by a tight array of associated proteins that do not appear to leave any unoccupied surface for processive movement of a kinesin (Yokoyama et al., 2004).

The defects in cilia motility reported here in Kif6 mutant mice may relate to the cilia motility defects observed after knockdown of Trypanosoma KIF9B, which disrupts beating and formation of the paraflagellar rod (PFR) within the flagellar membrane of T. brucei (Demonchy et al., 2009). However, since mammalian ependymal cilia do not have an analogous PFR structure, the precise location of Kif6 action within the vertebrate axoneme remains to be determined. One possibility is that Kif6 acts in the central pair, similar to the kinesin-9A proteins Kif9 in vertebrates and KLP1 in C. reinhardtii (Konjikusic et al., 2023; Han et al., 2022). This possibility is supported by recent mapping of the protein interactome in Tetrahymena thermophila by using cross-linking mass spectrometry, which identified links between Kif6 and the central-pair protein Spag16 and the nearby radial spoke-head protein Rsph4 (McCafferty et al., 2023 preprint). In fact, some redundancy between Kif6 and Kif9 might explain the finding that Kif6^p.G555fs ependymal cilia display decreased motility at D14 but recovered by D21 (Fig. 1J).

Kif6 mutation did not only affect cilia motility but also cilia polarity (Fig. 5), and these polarity defects are likely to contribute to the disturbed flow over LV wall tissue (Fig. 3F). Cilium orientation is regulated by apical microtubules (Vladar et al., 2012; Takagishi et al., 2020) and affected by apical actin networks between BBs (Werner et al., 2011), so Kif6 localization to this region is of interest as it is clearly distinct from that observed for Kif9 (Konjikusic et al., 2023). However, Kif6 mutation did not affect the overall appearance of apical microtubules (Fig. 5, Fig. S6), suggesting disturbed cilia polarity is secondary to the defects in cilia motility.

Indeed, cilia beating motility induces apical actin network organization that is crucial for BB stabilization (Mahuzier et al., 2018). Suppressed cilia motility would be expected to result in decreased apical actin networks around BBs, and destabilized BBs could be detached (Mahuzier et al., 2018). Because we observed all of these phenotypes it is possible that, despite the presence of Kif6 in the apical cytoplasm, Kif6-related actin defects are secondary to cilia-beating defects. A final possibility is that, because of BB destabilization, Kif6 mutant ependymal cilia with decreased motility are physically removed (by shedding), as suggested by Mahuzier et al., 2018 (Fig. 6D–F), and that stable cilia that remain until D21 display normal beating.

Kif6^pG555fs mutant mice were generated using the CRISPR-Cas9 system (Konjikusic et al., 2018). Kif6^pG555fs or C57Bl/6J mice were used for whole-mount staining, ependymal cell culture and LV wall tissue explant cultures. All mice were maintained at a 12/12 h light/dark cycle with no more than five mice per cage. Both male and female mice were used for the experiments, and sex differences were not observed. All mouse experiments were approved by the Animal Studies Committee at University of Texas at Austin (AUP-2021-00118).

Mouse Kif6 (mKif6) cDNA encoding the 803 aa protein (NP_796026) was cloned from a mouse brain cDNA library and sequenced. The cDNAs of full-length (FL) mKif6 (1-803aa) and motor-less (ML) mKif6 (360-803aa) were then subcloned into pEGFP-C1 vector (Addgene plasmid #54759) or pmCherry2-C1 vector (Addgene plasmid #54563; both a gift from Michael Davidson), yielding EGFP-Kif6 FL or mCherry-Kif6 FL or mCherry-Kif6 ML, respectively. mCherry was removed and mKif6 (1-493aa)-mNeonGreen was added into vector pmCherry2-C1 yielding mKif6 (1-493)-mNG.

The COS-7 (African green monkey kidney fibroblast-like) cell line was obtained from the American Type Culture Collection (ATCC; RRID: CVCL_0224), with cells cultured in DMEM (Gibco) with 10% (vol/vol) FetalClone III serum (HyClone) and 1% GlutaMAX supplement (Gibco) at 37°C under 5% CO₂. COS-7 cells were transfected with Trans-IT LT1 (Mirus Bio) according to the manufacturer's instructions.

COS-7 cells were collected 48 h post transfection. The cells were harvested by low-speed centrifugation at 1500 g for 5 min at 4°C. The pellet was rinsed once in PBS and resuspended in ice-cold lysis buffer (25 mM HEPES/KOH, 115 mM potassium acetate, 5 mM sodium acetate, 5 mM MgCl₂, 0.5 mM EGTA, and 1% Triton X-100 pH 7.4) freshly supplemented with 1 mM ATP, 1 mM phenylmethylsulfonyl fluoride, and protease inhibitor cocktail (#P8340; Sigma-Aldrich). After the lysate was clarified by centrifugation at 20,000 g for 10 min at 4°C, aliquots of the supernatant were snap-frozen in liquid nitrogen and stored at −80°C until further use.

HiLyte647-labeled microtubules were polymerized from purified tubulin including 10% Hily647-labeled tubulin (cytoskeleton) in BRB80 buffer (80 mM PIPES/KOH pH 6.8, 1 mM MgCl₂, and 1 mM EGTA) supplemented with 1 mM GTP and 2.5 mM MgCl₂ at 37°C for 30 min. 20 μM taxol in pre-warmed BRB80 buffer was added and incubated at 37°C for additional 30 min to stabilize microtubules. Microtubules were stored in the dark at room temperature (RT) for further use.

A flow cell (∼10 μl volume) was assembled by attaching a clean 1.5-mm thick coverslip (#1.5 Fisher Scientific) to a glass slide (Fisher Scientific) with two strips of double-sided tape. Polymerized microtubules were diluted in BRB80 buffer supplemented with 10 μM taxol, infused into flow cells and incubated for 5 min at RT for nonspecific adsorption to the coverslips. Subsequently, blocking buffer [15 mg/ml BSA in P12 buffer (12 mM PIPES/KOH pH 6.8, 1 mM MgCl₂, and 1 mM EGTA)] was infused and incubated for 5 min. Finally, KIF6 cell lysate in the motility mixture [2 mM ATP, 0.4 mg/ml casein, 6 mg/ml BSA, 10 μM taxol, and oxygen scavenger (1 mM DTT, 1 mM MgCl₂, 10 mM glucose, 0.2 mg/ml glucose oxidase, and 0.08 mg/ml catalase) in P12 buffer] was added to the flow cells. The flow cell was sealed with molten paraffin wax. To examine the movement direction of zebrafish and mouse Kif6 (dKif6 and mKif6, respectively), motility buffer containing the cell lysates of Kif6-mNG and KIF1A(1-393)-Halo^JFX554 was added into the flow cell, and the flow cell was sealed and imaged.

Images were acquired by total internal reflection fluorescence (TIRF) microscopy using an inverted microscope Ti-E/B equipped with the perfect focus system (Nikon), a 100×1.49 NA oil immersion TIRF objective (Nikon), three 20-mW diode lasers (488 nm, 561 nm and 640 nm) and an electron-multiplying charge-coupled device detector (iXon X3DU897; Andor Technology). Image acquisition was controlled using Nikon Elements software and all assays were performed at RT. Images were acquired, at a rate of one frame every 3 s for 3 min for mKif6, or at a rate of one frame every 100 ms for 10 s for KIF1A.

Maximum-intensity projections were generated, and the kymographs were produced by drawing along tracks of motors (width= three pixels) using Fiji/ImageJ2 . Motors frequently paused during motility events and, thus, velocity between pauses was analyzed. Velocity was defined as the distance of the kymograph (on the x-axis) divided by the time of the kymograph (on the y-axis).

LV wall tissues from three or four mouse brains at postnatal days 0 or 1 (P0 or P1, respectively) were collected and the suspended cells with trypsin-EDTA were cultured in DMEM with 10% FBS for 3 days in a laminin-coated flask. After shaking for 2 h, the adherent cells were plated on coverslips coated with poly-L-lysine and laminin, and differentiated into ciliated ependymal cells (ependymal cells) by serum starvation. Ependymal cells were then transfected with mKif6(1-493)-mNG or mCherry-Kif6 constructs using Lipofectamine 2000 (Invitrogen) 48 h before experiments.

For live imaging of Kif6(1-493)-mNG, D5 ependymal cells were transfected with Kif6(1-493)-mNG at D5 and imaged at D7 for 5 min in 0.2 frames per second (FPS), using a Ti2E inverted microscope with high-content screening system (Nikon). To observe cilia beatings, ependymal cells were plated on glass bottom dishes (20 mm, Cellvis). Time-lapse and digital image correlation (DIC) images were collected for 2 s at 47, 95 or 190 FPS by using a Ti2E microscope with a high-content screening system (Nikon). Ciliary tips were tracked using the manual tracking plug-in within Fiji for images shown in Fig. 3H and K. Kymographs shown in Fig. 3I were generated along the track of ciliary tips at 47 FPS for 200 ms. The duration of one beat (round-trip) of ciliary tips was measured in kymographs. Ciliary beating frequency (in Hz) was calculated as beats per second. Immotile cilia, as shown in Fig. 3I (bottom panel), were counted as 0.

LV wall tissues were collected from three mice each at P0, P4, P7 or P14. Each LV wall tissue or P45 testis tissue was lysed in PHEM buffer (50 mM PIPES, 50 mM HEPES, 1 mM EDTA, 2 mM MgSO4, 1% Triton X-100) with protease inhibitors, homogenized, and centrifuged for 1 h at 17800 g. The lysate supernatant was mixed with Laemmli sample buffer (Bio-Rad) and 2-mercaptoethanol and boiled. Samples were separated by SDS-PAGE and transferred to nitrocellulose membranes. The membranes were blocked with 2% BSA and probed with anti-Kif6 (1:1000, Proteintech, 17290-I-AP) or anti-GAPDH (1:1000, Cell Signaling Technology, #2118) antibodies. Detection was carried out using HRP-linked anti-Rabbit IgG (1:2000, Cell signaling Technology, #7074) and visualized using Pierce ECL substrate (Thermo Fisher Scientific).

Mouse brains were fixed in 4% paraformaldehyde at 4°C for overnight. The fixed whole brains were paraffin-embedded and cut into 5 µm sections for hematoxylin & eosin (H&E) staining. To measure cilia length, the fixed brains were sectioned in the coronal plane at 100 µm on a vibratome and immunostained. Whole-mount preparations of LV wall tissue were fixed with cold methanol for 10 min and 4% paraformaldehyde for overnight at 4°C. Cultured ependymal cells were fixed with cold methanol for 3 min and 4% paraformaldehyde for 10 min at RT. Whole-mount, ependymal cell or vibratome-sectioned samples were blocked with 10% goat serum in PBS for 30 min and incubated with primary antibodies for overnight at 4°C. Primary antibodies, were anti-Kif6 (1:100, Proteintech, 17290-I-AP), anti-Arl13b (1:200, Proteintech, 17711-I-AP), anti-Acetylated Tubulin (1:200, Sigma, T7451), anti-gamma Tubulin (1:100, Abcam, ab11321), anti-beta Tubulin (1:100, Abcam, ab7287), anti-Centrin (1:50, Sigma, 04-1624), anti-FGFR1OP (1:100, Proteintech, 11343-1-AP), anti-Dnai1 (1:100, NeuroMab, 73-372), or anti-Tubulin [YL 1/2] (1:200, Abcam, ab6160), were detected with secondary antibodies conjugated to Alexa Fluorophores (Invitrogen). The apical surface of ependymal cells was imaged using a confocal laser scanning microscope LSM700 (Zeiss) or LSM 980 Airyscan 2 SR mode with joint Deconvolution (Nikon) for super-resolution images. H&E or immunostained tissue sections were imaged using a BZ-X710 fluorescence microscope (KEYENCE).

The lateral sides of LV wall tissue were dissected from mouse brain at P12 and pinned on a silicon plate covered with DMEM. Fluorescent microsphere (1 µm) -containing PBS was placed on the middle part of LV walls. Bead migration was recorded in ten FPS for 10 s using an AXIO Zoom.V16 microscope (Zeiss). Time-projection images were created from 100 frames obtained by using Fiji software (Schindelin et al., 2012). Beads migration speed was calculated by using the manual tracking plug-in of Fiji.

Mouse LV wall tissues obtained at P14 were fixed overnight using 3% glutaraldehyde with 2% paraformaldehyde. Tissues were post fixed in 1% osmium tetroxide with 1% potassium ferrocyanide for 3 h. After en-bloc staining with 1% aqueous uranyl acetate for 1 h, ethanol-dehydrated tissues were displaced into aceton and embedded in epoxy resin. Tissues were sectioned (80 nm thickness) with an UltraCut Ultramicrotome (Leica). Images of the apical cell cortex of ependymal cells were obtained using a Tecnai Transmission Electron Microscope (FEI).

The violin plots were generated using GraphPad Prism 9 (GraphPad) Welch's t-test or Mann–Whitney U-test were used to compare the means of two experimental groups by using GraphPad Prism. Differences were considered statistically significant at P<0.05. To determine the rotational BB orientation for a single ependymal cell (Fig. 5E), a vector was drawn from the center of a FOP dot to the center of the closest γ-tubulin dot in each cell; more than ten vectors were averaged. Vector angles were measured using Fiji and plotted as a circular diagram using the statistical software R (R Core Team, 2019: https://www.r-project.org/). Mean resultant vector lengths (⁠⁠) were calculated using the ‘circular’ package (https://cran.r-project.org/web/packages/circular/index.html) within the R statistical computing environment. For Fig. 5I, vector angle distribution of cilia bundles on WT and Kif6^pG555fs LV wall tissue was compared by using Watson's two-sample U2 test (R software).

Kif6 mice were generated by Mia J. Konjikusic (University of California, San Francisco) and overseen by Elle C. Roberson (CU Anschutz, Colorado). Live imaging of cilia beatings and super-resolution imaging were performed at the Microscopy and Imaging Facility at UT Austin (RRID# SCR_021756).