The kinesin-4 protein KIF27 forms a cytoskeletal scaffold at the transition zone to promote motile cilia structural integrity

Significance

Motile cilia are chemo- and mechanosensory organelles essential to the respiratory, nervous, and reproductive systems. Although prior research has identified ciliary components and studied their atomic-level structures, we still do not know how motile cilia are formed in vertebrates. Our innovative approaches identify a kinesin KIF27-mediated cytoskeletal scaffold in regulating the barrier functions of the transition zone (TZ) to drive mammalian motile cilia assembly. Such function of KIF27 is distinct from the function of its paralog, KIF7, in regulating Hedgehog signaling in primary cilia. Results demonstrate that proper organization of the TZ is essential for mammalian motile cilia assembly in vivo and that the TZ’s barrier function dictates the cilium’s protein composition pertinent to cilia motility.

Abstract

Motile cilia are eukaryotic organelles with essential chemo- and mechanosensing functions across evolution, from single cell organisms to humans. Motile cilia of the mammalian nervous, respiratory, and reproductive systems are characterized by unique motility proteins to generate fluid flow essential for transporting metabolites and removing mucus. The molecular mechanism underlying motile cilia assembly remains unknown. Here, we use high-resolution imaging, proteomics, in situ cryotomography, and single-molecule motility assays to identify mammalian KIF27, a motor protein of the Kinesin-4 family and homologue of the Hedgehog pathway regulator COS2/KIF7, as a key regulator of motile cilia assembly in vivo. We show that KIF27 promotes the integrity of the transition zone (TZ), a diffusion barrier situated at the cilium base. Loss of KIF27 results in specific and profound defects in axonemal structure and disrupts cilia beating, which collectively lead to organismal phenotypes that recapitulate primary ciliary dyskinesia (PCD). We show that the motile properties of KIF27 are dispensable for its function in motile cilia assembly. Instead, KIF27 acts as a microtubule scaffold to regulate the TZ architecture and enable correct ciliary incorporation of motility-generating proteins. Given that KIF27 orthologues exist in different evolutionarily lineages, we propose that the ancestral activity of KIF27/KIF7 kinesins was to form a microtubule-associated scaffold for protein–protein interactions pertinent to cilia formation and signaling. The transition-zone-associated KIF27 activities may represent a general building principle for the motile cilia assembly in diverse species and cell types.

Primary cilia and motile cilia are microtubule-based organelles found in all modern eukaryotes and have diverse mechanosensory functions. The primary cilium is well recognized as a hub of signaling pathways, including the hedgehog pathway (1, 2). Motile cilia, in contrast, can produce synchronized waveforms and drive fluid flow to regulate locomotion of marine larvae and transport mucus and metabolites across brain ventricles, respiratory and auditory epithelia, and reproductive tracks in mammals (3, 4). At the respiratory mucosal barrier, motile ciliated cells constitute the foremost host defense mechanism by generating coordinated ciliary flow for mucociliary clearance of attached pathogens (3, 5). Dysfunctional primary and motile cilia can lead to ciliopathies and motile ciliopathies, respectively, characterized by nonoverlapping clinical presentations (3, 6). Primary ciliary dyskinesia (PCD) is a form of motile ciliopathy caused by genetic mutations that impair cilia beating. PCD patients exhibit chronic respiratory and ear inflammation, while some also experience situs inversus and subfertility (3, 5, 7). Despite the importance of motile cilia to human physiology and disease, the molecular mechanisms that regulate motile cilia biogenesis and assembly are not well understood.

Both primary and motile cilia are compartmentalized from the cytosolic environment by unique membrane and cytoskeletal compositions to optimize for the specialized ciliary functions. Within cilia, axonemal microtubule doublets of a ninefold symmetry are templated from modified centrioles, the basal bodies, which are docked onto the plasma membrane. During ciliogenesis, tubulin heterodimers, signaling molecules, and various ciliary structural components are transported by the intraflagellar trafficking (IFT) complex attached to the anterograde and retrograde motor proteins (8, 9). At the base of the cilia sits the transition zone (TZ), a suborganellar domain composed of conserved proteins, including CEP290, the Nephronophthisis (NPHP) complex, and the Meckel–Gruber syndrome (MKS) complex (10, 11). The organization of mammalian TZ in the primary cilium is well defined, and genetic studies demonstrate a gating mechanism for compartmentalization, in which the TZ regulates ciliary cargo entry and exit by acting as a critical diffusion barrier (12). Many human ciliopathies are caused by mutations that perturb TZ functions associated with the primary cilium. Motile cilia contain, in addition to the axonemal building blocks and TZ proteins, macromolecular machineries to drive coordinated cilia beating. These motility components include the central pair (CP) apparatus, radial spokes (RS), outer and inner dynein arms, as well as a growing list of microtubule inner proteins (MIPs) (8, 13, 14), which occupy the luminal space of axonemal microtubules and function to stabilize the axoneme and regulate motility (13, 15, 16). Much of our understanding of motile cilia comes from studies in basal eukaryotes, such as Chlamydomonas and Trypanosome brucei. The molecular mechanisms that regulate motile cilia assembly and function in mammals have not been defined. It remains unknown how these motility proteins are incorporated into the ciliary compartment during motile ciliogenesis, largely due to limited mammalian organismal and cell biological models for studying motile cilia.

To uncover the molecular mechanisms governing mammalian motile cilia assembly, we employed an in silico strategy to identify candidate components preferentially associated with multiciliated cells. By combining high-resolution cilia imaging, cilia proteomics, in situ cryotomography, and in vitro microtubule assays, we found that correct assembly of motile cilia depends on KIF27, a motor protein of the kinesin-4 family and paralogue to the Hedgehog pathway regulator COS2/KIF7 (17–22). Loss of KIF27 in mice causes specific and profound defects in axonemal structure and disrupts cilia beating, which collectively lead to organismal phenotypes that recapitulate PCD. Unexpectedly, KIF27 does not use its motor function for motile cilia assembly. Instead, our data suggest a model in which KIF27 acts as a microtubule-associated scaffold at the TZ to promote ciliary incorporation of motility proteins and the structural integrity of the ciliary axoneme. Given that KIF27 orthologues are present in many organisms with motile cilia, the TZ associated KIF27 activities may represent a general building principle for motile cilia assembly in diverse species and cell types.

Results

In Silico Enrichment for Conserved Motile Cilia-Associated Genes.

Motile cilia possess unique machineries to generate the “9 + 2” axonemal structure and motility (Fig. 1A). To uncover regulators underlying motile cilia assembly, we carried out a stepwise in silico curation to select for evolutionarily conserved genes specifically associated with motile ciliogenesis (Fig. 1B). Using publicly available single-cell RNA sequencing datasets, we first compiled a list of differentially expressed genes from motile ciliated cells of the mouse and human airways (23), mouse ependyma (24), as well as human fallopian tubes (25). Next, we removed entries from ciliopathy and PCD gene panels (3, 6), as well as entries that encode proteins identified by proteomics from both primary and motile cilia (26). Our filtering strategy led to about 80 genes (SI Appendix, Table S1), including MAP9 (27), STK33 (28, 29), KIF9 (30, 31), and TTLL9 (32), which are implicated in different aspects of motile cilia formation and function. From this list, we then performed text mining to search for entries with reported motile cilia phenotypes and prioritized those with orthologues in ciliated species ranging from Trypanosoma, sea anemone, to mice and humans (Fig. 1B). One of the entries that met all the criteria was Kif27, a motor protein that belongs to the Kinesin-4 family and is homologous to the Hh pathway regulators Cos2 (33–35)/Kif7 (22, 36) (Fig. 1C). siRNA targeting Planaria Kif27 disrupted ciliogenesis and locomotion (36). A lacZ insertion to the mouse Kif27 locus led to early lethality with hydrocephalus and nasal mucus accumulation (37). Ectopically expressed KIF27-GFP localized to the base of motile cilia (38–40). These organismal defects and expression patterns suggest an evolutionarily conserved yet uncharacterized role for Kif27 in motile cilia biogenesis.

Fig. 1.

KIF27 is required for mammalian motile ciliogenesis. (A) Central pair (CP) microtubules, dynein regulatory complex (DRC), and radial spokes (RS) are present in “9 + 2” motile cilia and absent from “9 + 0” primary cilia. The transition zone (TZ) sits at cilia base. (B) in silico curation for conserved motile cilia genes. DEG: differentially expressed genes. (C) Phylogenetic tree displaying Kif27 homologs in ciliated organisms and respective domain organizations and RefSeq IDs; kinesin motor (blue) and coiled-coils (yellow). (D) Phenotypic characterizations of postnatal Kif27^−/− brains (Upper panels). Ventricular zone (v), hippocampus (hc), olfactory bulb (ob). Loss of corpus callosum (arrowhead) and enlarged lateral ventricle (arrow) were indicated. Mucus accumulation was present in Kif27^−/− nasal cavity (arrowheads) (Lower panels). (Scale bar, 2 mm.) (E) TEM images of p17 wt and Kif27^−/− respiratory cilia; numbers indicate axonemal organizations. Arrows indicate singlet CP microtubules. Arrowheads indicate peripheral doublets. (Scale bar, 500 nm.) (F) Quantification of CP organizations of respiratory cilia (wt = 114, Kif27^−/− = 204, N = 4). Others include “9 + 1,” “9 + 0,” “8 + 0,” and disorganized axonemes. (G) Expansion microscopy images of p17 wt and Kif27^−/− cilia labeled with acetylated α-tubulin (magenta). (Scale bar, 500 nm.) (H) Expansion microscopy images of p17 wt and Kif27^−/− cilia labeled with acetylated α-tubulin (magenta) and SPAG6 (cyan). Fluorescence intensities of SPAG6 and acetylated α-tubulin of indicated axonemes (arrows) for wt and mutants (Right panel). (Scale bar, 2 µm.) (I) Quantification of CP and SPAG6 organizations in postnatal cilia (n = 260 for wt and Kif27^−/− cilia, respectively, N = 3). (J) Expressions of RSPH9 and DNALI1 (cyan) in expanded postnatal wt and Kif27^−/− cilia. Acetylated α-tubulin (magenta) marks axonemal microtubules. Scale bar = 1 µm for sideview and Scale bar = 500 nm for cross-section insets. Error bars indicate SD.

KIF27 Is Required for Motile Cilia Assembly, But Dispensable for Hh Signaling During Embryogenesis.

To investigate KIF27 function and molecular mechanism in mammalian motile cilia, we generated a Kif27 knockout (Kif27^−/−) mouse line by removing the first protein coding exon of Kif27 (SI Appendix, Fig. S1 A and B). All Kif27^−/− homozygous null mice we examined developed prominent domed-shape heads, nasal mucus obstruction associated with early postnatal lethality with an average life span of 2 wk; male null mutants displayed defects in spermatogenesis (Fig. 1D and SI Appendix, FigS. S1C and S9 A and B). Using scanning electron microscopy (SEM) and immunostaining, we observed abnormalities in the brain ependyma and morphological defects for respiratory motile cilia including reduced cilia length, supporting a role for KIF27 in motile cilia assembly (SI Appendix, Fig. S2 A–C).

KIF27 is a paralogue of mammalian KIF7, which plays a dual role in Hh signaling by regulating primary cilia structure and controlling primary ciliary dynamics of the GLI transcription factor, the final effector of Hh signal transduction (33–35, 41). In contrast to Kif7^−/− embryos which show gain-of-function Hh mediated patterning defects, Kif27^−/− mice did not show detectable defects in primary cilia-mediated Hh-dependent developmental processes (SI Appendix, Fig. S3 A–C). Kif27^−/−; Kif7^−/− double mutants died at birth and exhibited polydactyly (SI Appendix, Fig. S3B), phenocopying the Kif7^−/− single mutant defects. Kidney cysts, which can be caused by primary cilia defects (42), were not present in postnatal Kif27^−/− mice (SI Appendix, Fig. S3E). These genetic and phenotypic analyses provide evidence against a strong requirement for Kif27 in primary cilia-mediated Hh signaling and kidney morphogenesis during embryonic development.

Loss of KIF27 Causes Defects in Motile Cilia Axonemal Structure.

Using transmission electron microscopy (TEM), we observed an array of axoneme abnormalities in postnatal Kif27^−/− respiratory cilia. The normal “9 + 2” configuration of motile cilia with 9 outer doublets + 2 singlet CP microtubules was only present in < 20% of Kif27^−/− respiratory cilia (Fig. 1E). Instead, a fraction of axonemes showed either single CP microtubules or two CP microtubules without the bridge; nearly 40% of cilia completely lacked CP microtubules (9 + 0); about 30% of Kif27^−/− cilia displayed < 9 outer doublets, open B-tubules, and disorganized doublets (Fig. 1F). Consistent with TEM analysis, tissue expansion microscopy with acetylated α-tubulin staining revealed striking cilia defects, including missing CP, reduction in outer doublets, as well as misarranged doublets in respiratory epithelia of postnatal Kif27^−/− mice (Fig. 1G).

We next used expansion microscopy to characterize cilia defects in Kif27^−/− mutants in more detail. SPAG6 is a CP apparatus protein associated with one of the CP singlet microtubules (43). In both embryonic and postnatal Kif27^−/− mice, > 70% reduction in SPAG6+ motile cilia were observed compared to wild-type (wt), suggesting that axonemal defects arise early in Kif27^−/− mice (Fig. 1 H and I and SI Appendix, Fig. S2 D–I). In wt respiratory cilia, SPAG6 signal appeared as discrete puncta adjacent to the CP microtubule labeled with acetylated α-tubulin (Fig. 1H and SI Appendix, Fig. S2H). About 50% of Kif27^−/− cilia lacked detectable signals for both SPAG6 and CP tubulin (SPAG6- CP-); about 25% of Kif27^−/− cilia were SPAG6- CP+; 2% of Kif27^−/− cilia were categorized as SPAG6+ CP-, suggesting that SPAG6 is unlikely to be recruited to motile cilia in the absence of KIF27 (Fig. 1I and SI Appendix, Fig. S2I). In contrast, RSPH9 (44) and DNALI1 (45), which are subunits of RS and the inner dynein arm complex, respectively, were organized in the ciliary compartment similarly between wt and Kif27^−/− cilia, regardless of the presence of CP (Fig. 1J and SI Appendix, Fig. S2J).

Dynamic Localization of KIF27 During Motile Ciliogenesis.

To characterize the endogenous KIF27 localization, we generated a KIF27-GFP reporter allele in which GFP was fused to the last exon of the mouse Kif27 locus (SI Appendix, Fig. S4A). The homozygous Kif27^GFP/GFP as well as Kif27^GFP/- mice were viable and fertile, indicating that the KIF27-GFP fusion protein is functional and can inform the endogenous localization of KIF27 during different stages of motile ciliogenesis. In the embryonic respiratory ciliated cells of Kif27^GFP/GFP animals, GFP signal was first detected in the deuterosome stage and colocalized with CP110 (46, 47), a centrosomal protein, at the distal end of the amplifying centrioles prior to cilia formation (Fig. 2A). When centrioles were docked onto the apical plasma membrane, GFP signals colocalized with CEP164 (48), a cilia transition fiber marker (Fig. 2B). When ciliation was initiated, KIF27-GFP was enriched in the axonemal distal ends of newly formed short cilia and became less abundant in long and mature cilia (Fig. 2C). In contrast, KIF27-GFP was not observed in the primary cilium (SI Appendix, Fig. S3D), consistent with a motile-cilia-specific role for KIF27. The dynamic association of KIF27-GFP with centrioles and nascent ciliary axoneme suggests that KIF27 functions may be temporally regulated during ciliogenesis.

Fig. 2.

KIF27 localizes to motile cilia and interacts with CEP162. (A) KIF27-GFP (cyan), CP110 (magenta), and SAS6 (yellow) expressions at the deuterosome stage of E16 Kif27^GFP/GFP nasal epithelium. (B) KIF27-GFP (cyan) and CEP164 (yellow) expressions during centriole amplification and docking. Acetylated α-tubulin (magenta) labels centriolar microtubules. (C) KIF27-GFP (cyan) expressions during ciliogenesis. Acetylated α-tubulin (magenta) marks axonemal microtubules. The arrowhead represents ciliary tip, and arrows represent centriole distal tip/TZ. DAPI (gray) labels the nucleus. [Scale bar, 2 µm for (A–C).] (D) Expansion microscopy images of KIF27-GFP (cyan) and CEP162 (yellow) during centriole amplifications; fluorescent intensities shown on the Right panel. Acetylated α-tubulin (magenta) labels centriolar microtubules. (Scale bar, 500 nm.) (E) Expressions for KIF27 (cyan), labeled with an anti-KIF27 antibody, and CEP162 (yellow) at the TZ (arrows) and cilia tip (asterisks) in the embryonic wt respiratory cilia. (Scale bar, 2 µm.) (F) Co-IP of KIF27-GFP and CEP162 by a GFP nanobody from Kif27^GFP/GFP mouse testis and not from wt (Kif27 ^+/+). Input (cell lysate) and beads (eluate fraction) were blotted with anti-GFP and anti-CEP162 antibodies. (G and H) Left: Domain architectures of full-length human KIF27 (G) and full-length human CEP162 (H) and truncation constructs [(i)–(iv)]. Right: anti-HA immunoprecipitation (IP) from HEK293T cells. Input (cell lysate), beads (eluate fraction) and IgG (eluate fraction from IgG isotype control) were blotted anti-HA and anti-GFP antibodies. Positive and negative Co-IPs were indicated as “+” and “–”. (I and J) Anti-HA and anti-CEP290 coimmunoprecipitation of full-length KIF27, CEP162, and CEP290 overexpressed in HEK293T cells. Input (cell lysate), beads (eluate fraction) and IgG (eluate fraction from IgG isotype control) were subject to immunoblotting with anti-FLAG, anti-HA, and anti-GFP antibodies. (K) Schematic illustrating the 3-way complex model for CEP290-CEP162-KIF27. CC: Coiled coil, SMC: SMC homology domain, Myosin-tail: myosin-tail homology domain.

KIF27 Associates With TZ Protein CEP162 in Motile Ciliated Cells.

In human U2OS cells without motile cilia, CP110 was identified as a component of the centriole distal-end complex together with CEP97, CEP290, and CEP162 (49). CEP162, which binds to CEP290, is a microtubule-associated TZ protein (50). Consistent with the colocalization of KIF27-GFP with CP110, we observed that in Kif27^GFP/GFP mouse embryonic motile ciliated cells, KIF27-GFP colocalized with CEP162 at the distal end of amplifying centrioles (Fig. 2D). To characterize the subcellular localizations of endogenous untagged KIF27, we generated an antibody that binds to the central region of KIF27 (SI Appendix, Fig. S1 D–G). Consistent with our KIF27-GFP analyses, antibody-labeled KIF27 in wt respiratory epithelial cells specifically colocalized with CEP162 at the distal end of centrioles as well as the TZ in newly formed motile cilia, in which KIF27 was also enriched at the cilia tip (Fig. 2E).

To determine whether KIF27 can form a protein complex with CEP162, we purified a GFP nanobody to immunoprecipitate KIF27-GFP from mouse multiciliated tissues, including the testis, which express KIF27 (40) (SI Appendix, Fig. S4 B and C). In testis lysates from adult Kif27^GFP/GFPanimals, endogenous CEP162 coimmunoprecipitated with KIF27-GFP (Fig. 2F). To locate the interaction domains in each protein, structure–function analyses were performed by coexpressing truncation constructs of KIF27 and CEP162 (50) (Fig. 2 G and H). The binding domains were mapped to the C-terminus of KIF27 (KIF27-CC2) and the N-terminal flexible region of CEP162 (CEP162-N). In vitro pull-down assays using purified proteins confirmed that KIF27-CC2 directly interacts with CEP162-N (SI Appendix, Fig. S5 A–C). To visualize the KIF27–CEP162 interaction in a cellular context, we expressed KIF27-GFP and HA–CEP162 binding constructs in U2OS cells. When expressed alone, KIF27-CC2 appeared in dispersed intracellular puncta, whereas CEP162-N showed a strong nuclear localization. In cells that coexpressed KIF27-CC2 and CEP162-N, both proteins showed translocation to cytoplasmic microtubule bundles labeled with acetylated α-tubulin, a marker for stable microtubules (SI Appendix, Fig. S5 D and E).

CEP162 can interact with CEP290, a core component of the TZ complex (50). Consistent with the literature, coimmunoprecipitations between these two proteins were observed when full-length CEP162-HA and CEP290-FLAG were heterologously expressed. In contrast, no detectable coimmunoprecipitation was observed for full-length KIF27-GFP and CEP290-FLAG (Fig. 2I). Given that the KIF27-binding domain of CEP162 is different from its microtubule-binding and CEP290-binding domains, which are located to the central coiled coils of CEP162 (50), we tested whether CEP162 can couple KIF27 to CEP290 to form a 3-way complex. Coimmunoprecipitation between CEP290 and KIF27 was only observed when CEP290, KIF27, and CEP162 were heterologously expressed together (Fig. 2 J and K). Given that CEP290 can form a bridge between the cilia membrane and microtubule-based axoneme, the results suggest that a physical association between KIF27 and CEP162 can promote the recruitment of KIF27 to the motile cilia TZ that contains CEP290 as well as the association of KIF27 with TZ microtubules.

KIF27 Fine-Tunes the Recruitment of Ciliary TZ Proteins.

CEP162 can promote TZ assembly in human retinal pigmented epithelial cells and is implicated in retina degeneration, a defect attributed to primary cilia dysfunction (51). CEP290 plays an evolutionarily conserved role in regulating TZ assembly and is implicated in a wide spectrum of human ciliopathies⁴⁶. Given that KIF27 interacts with the CEP162–CEP290 complex and colocalize to the TZ, we tested whether KIF27 can regulate TZ assembly in motile cilia. We first examined the suborganellar architecture of the mammalian motile cilia TZ. Using lattice SIM superresolution microscopy, we localized the TZ proteins CEP164, CEP290, MKS1, MKS3, and RPGRIP1L in respiratory ciliated epithelial cells from wt and Kif27^−/− mice (Fig. 3A). In postnatal motile ciliated cells, the proximal–distal distances were measured for these markers relative to CEP164 to reflect the suborganellar architecture of motile cilia TZ. The spatial organization for these ciliary TZ proteins in motile cilia was not identical to those observed for the primary cilium (52, 53) (Fig. 3B). In both wt and Kif27^−/− cilia, these TZ proteins were organized in similar patterns in which MKS1 was the closest to CEP164, CEP290, and RPGRIP1L overlapped at a region distal to MSK1, and MKS3 was most distal to CEP164 (SI Appendix, Fig. S6A). Using conventional confocal microscopy, we then measured the fluorescence intensities for these TZ proteins in postnatal respiratory motile cilia. Signal intensities for CEP290, RPGRIP1L, and MKS1 were significantly reduced in Kif27^−/− motile cilia, while MSK3 was unaffected (Fig. 3C). The data indicate KIF27 is required to recruit the correct amount of TZ proteins to motile cilia.

Fig. 3.

KIF27 plays a role in regulating motile cilia TZ assembly. (A) Lattice SIM2 images of TZ markers (all in cyan), with γ-tubulin (yellow) from p17 cilia. (Scale bar, 500 nm.) (B) A model for mammalian motile cilia TZ. (C) Fluorescence intensities quantified using LSM980 for TZ markers from p17 cilia. CEP164 (wt = 224, Kif27^−/− = 231), MKS1 (wt = 145, Kif27^−/− = 137), CEP290 (wt = 127, Kif27^−/− = 120), RPGRIP1L (wt = 138, Kif27^−/− = 134), MKS3 (wt = 134, Kif27^−/− = 130). N = 6 for CEP164 and N = 4 for the other markers. (D) Expansion microscopy images of CEP290 and RPGRIP1L (cyan), acetylated α-tubulin (magenta), and CEP162 (yellow) from embryonic cilia with a cartoon for TZ organization on the Right. (Scale bar, 2 µm.) (E) Average TZ height (h) and (F) TZ height differences for CEP290 and RPGRIP1L from embryonic wt and Kif27^−/− cilia. For (E and F), images shown in (D) were used for measurements, n = 90 for each genotype, N = 3. Plots show 10 to 90% for (E and F). *P < 0.05, **P < 0.005, ***P < 0.0005, ****P < 0.0001, ns, not significant. (G) En face images of CEP290 (cyan) for embryonic cilia. (Scale bar, 1 µm.) (H) Quantification of CEP290 puncta (n = 90, N = 3 for each genotype). Error bars represent SD.

To investigate how KIF27 regulates TZ organization in greater detail, we used expansion microscopy to resolve the morphological nuances of TZ proteins. In ciliated respiratory cells isolated from wt mice, RPGRIP1L and CEP290 were enriched in a domain distal to CEP162 located at the base of cilia (Fig. 3D). From a lateral view, RPGRIP1L and CEP290 occupied a broader space than CEP164 and CEP162 (Fig. 3D and SI Appendix, Fig. S6B). In Kif27^−/− respiratory cilia, RPGRIP1L and CEP290 domains were distally expanded compared to wt and exhibited asymmetrical distribution along the axoneme (Fig. 3 D–F). TZ markers in motile cilia displayed a ring-like morphology composed of discrete puncta with a ninefold symmetry, similar to TZ markers in primary cilia (54) (Fig. 3G and SI Appendix, Fig. S6C). In Kif27^−/− motile cilia, gaps between puncta were more frequently present in TZ rings of CEP162 and RPGRIP1L (Fig. 3G and SI Appendix, Fig. S6C). The CEP290 TZ rings in Kif27^−/− cilia often displayed < 9 puncta, and >60% of the rings showed 8 or less-than 8 puncta (Fig. 3H). The data support an important role for KIF27 in regulating proper TZ assembly in motile cilia.

KIF27 Regulates Motile Ciliary Protein Composition.

We hypothesized that an abnormally organized ciliary TZ may affect the ciliary recruitment of motility proteins, which may account for the structural and beating defects of motile cilia observed in Kif27^−/− mutants. In whole cell lysates collected from wt and Kif27^−/− mouse tracheal epithelial cells (mTECs), comparable protein expressions for subunits of dynein arm complex, CP apparatus, MIPs, and IFT complex were detected (Fig. 4A), indicating that these motility proteins were not degraded. To test whether ciliary recruitment of motility proteins is regulated by KIF27, we determined the motile cilia proteome using mass spectrometry. Cilia fractions from mTECs derived from wt and Kif27^−/− respiratory epithelia were isolated using a high-salt induced deciliation protocol (Fig. 4B). The protein abundance was obtained from three biological replicates for each genotype, and the abundancy ratio (WT/Kif27 KO) was normalized with an outer dynein arm component DNAH5, given that dynein outer arms were present in most Kif27^−/− cilia based on TEM analysis (Fig. 1E). Proteomics from three replicates consistently showed that essentially all components of the CP apparatus were severely depleted from the Kif27^−/− cilia fraction (Fig. 4C). Unexpectedly, the outer doublets–associated MIPs, including CFAP20, Tektin, and PIERCE complexes (16), were also significantly reduced in the Kif27^−/− ciliary compartment (Fig. 4C). The abundance of dynein arms and RS proteins were not severely perturbed in mutant cilia. In contrast, core components of IFT components and the Bardet–Biedl syndrome protein complex (BBSome) were mildly upregulated in the Kif27^−/− cilia fraction. Conserved signaling modules, membrane proteins, and microtubule associated proteins were differentially affected by the lack of KIF27 (Fig. 4C). Abnormal protein composition observed in Kif27^−/− cilia fraction indicates impaired barrier functions for the motile cilia TZ in the absence of KIF27, and that TZ dysfunction may account for the missing structural proteins and impaired cilia motility observed in Kif27^−/− cilia.

Fig. 4.

KIF27 is required for proper barrier functions for motile cilia TZ. (A) Total cell lysates of wt and Kif27^−/− mouse tracheal epithelial cells (mTEC) subjected to immunoblotting. Protein expression levels of ciliary proteins are unaffected in Kif27^−/− cells. Acetylated α-tub marks cilia and β-actin serves as loading control. (B) Schematic representation of proteomic analysis following cilia isolation. (C) Heat map of abundance levels normalized for dynein axonemal heavy chain 5 (DNAH5) in cilia isolated from wt and Kif27^−/− mTEC (three replicates for each genotype) for representative proteins. Colors indicate fold change. MIPs: microtubule inner proteins; RS: radial spoke proteins; ODR: outer dynein complex; IDR: inner dynein complex.

KIF27 Promotes Ciliary Incorporation of Microtubule Inner Proteins.

To examine the spatial morphology of motile cilia axonemes in more detail, we employed in situ cryoelectron tomography (Cryo-ET) to reveal the native arrangements of axonemal microtubules and associated structural proteins. Vitrified motile ciliated cells dissociated from wt and Kif27^−/− respiratory epithelia and mTECs were first identified based on the expression of Centrin-GFP with correlative light and electron microscopy (Fig. 5 A and B). These cells were thinned with a focused ion beam (FIB) and imaged by cryo-ET (SI Appendix, Fig. S7 a-e). Under the native cellular environment, microtubule outer doublets, CP microtubules, and CP apparatus were clearly visible in wt cilia (Fig. 5C). In Kif27^−/− motile cilia, both CP+ and CP- cilia were observed, consistent with our TEM and expansion microscopy analysis (Fig. 5 D and E). In those Kif27^−/− mutant cilia with CP microtubules, densities correlated with the CP apparatus proteins appeared less organized spatially compared to the wt cilia. To examine whether MIPs remain associated with the microtubule doublet, we performed subtomogram averaging. Consistent with proteomics analysis, densities of the Tektin complex as well as NME7 were reduced in the outer microtubule doublets in Kif27^−/− cilia (Fig. 5 F and G). Regardless of the presence of CP, a conspicuous number of extra densities were present within the luminal space of Kif27^−/− mutant cilia and masked the densities of axonemal structural components (Fig. 5 C–E and SI Appendix, Fig. S8). The extra density only appeared in Kif27^−/− mutant cilia, which indicates that it is not part of the normal axonemal assembly and is consistent with ectopic retentions of proteins in the mutant cilia observed from cilia proteome. The data are consistent with a key role for KIF27 in motile cilia assembly by promoting the recruitment and correct incorporation of motility proteins into the ciliary compartment.

Fig. 5.

In situ cryo-ET characterization of Kif27^−/− cilia. (A and B) The cilia array with clear Centrin-GFP signal indicates respiratory ciliated cell, which were selected for further FIB milling. (Scale bar, 50 μm.) (C and D) Representative tomographic slices of wt and Kif27^−/− cilia. wt cilia with CP microtubules (green arrowhead) with clear background in the ciliary luminal region (C); Kif27^−/− cilia showed extra protein density between CP and microtubule outer doublets (D), Bottom: cross-sections, slice thickness 193.36 Å. CM: ciliary membrane, MTDs: microtubule doublets, CA: central apparatus. (Scale bar, 100 nm.) (E) A tomogram slice of representative Kif27^−/− cilia lacked CP; CP microtubules are absent from the lumen (green arrowhead); a decrease in MTDs numbers (from 9 to 8) with altered doublet alignment (highlighted as 5 in red). Top: longitudinal central slices along cilia, slice thickness 48.34 Å; (Scale bar, 100 nm.) (F and G) The cross-sections of MTDs subtomogram averaging maps highlighted the densities for tektins (green) and NME7 (blue) in wt (F) and Kif27^−/− cilia (G). The subtomogram averaging maps were obtained from the 8-nm repeat distance segments. The local resolution filtered maps are shown.

Phenotypical Heterogeneity Associated With Brain Ependymal and Oviductal Cilia.

To investigate whether KIF27 is required for motile cilia biogenesis in different cell types, we examined the cellular defects associated with brain ependymal cilia and fallopian tube cilia, which also express KIF27 (39). Consistent with histology and SEM analysis, we observed significant reduction in cilia length and CFAP20 fluorescence intensity in Kif27^−/− brain ependymal cilia (SI Appendix, Fig. S9 C–F). In contrast, oviductal cilia from the fallopian tube of Kif27^−/− female mice exhibited normal length with comparable CFAP20 expression along the ciliary axoneme (SI Appendix, Fig. S10 A–D). Using live imaging, we observed similar beating frequencies for oviductal cilia from Kif27^−/− mutants and wt littermates (SI Appendix, Fig. S10H and Movies S5 and S6). We next employed tissue expansion microscopy to resolve structural nuances in CP and TZ organization. Similar to respiratory cilia, CEP290 and RPGRIP1L were localized to the ciliary base in both wt and Kif27^−/− oviductal cilia (SI Appendix, Fig. S10I). In contrast, only 25% of Kif27^−/− oviductal cilia exhibited detectable SPAG6 and CP microtubules (SI Appendix, Fig. S10 E–G). The data suggest that motile cilia from different multiciliated cells may exhibit tissue-specific heterogeneity, and that activities for KIF27 in these tissues may be differentially regulated to promote structural integrity.

KIF27 Exhibits ATP Dependent Motility Toward Motile Cilia-Specific Microtubules.

Some members of the kinesin-4 family, for instance KIF4 (55) and KIF21 (56, 57), show ATP dependent microtubule plus-end directed motility, while others, such as KIF7, lack ATP-mediated translocations along microtubules (35). The ATP-dependent translocation ability of KIF27 was reported with a chimeric KIF27 construct, which contains the KIF27 motor domain and a leucine zipper for dimerization (58), raising the possibility that motility may be important for KIF27-mediated cilia assembly. To address the molecular mechanism of KIF27 in motile cilia, we first characterized the biochemical properties of purified KIF27 motor proteins using well-established TIRF-microscopy assays (35, 59) (Fig. 6 A and B). A GFP tagged truncated KIF27 motor dimer, including the N-terminal motor domain and the first coiled-coil of KIF27, KIF27_1-570-GFP, was purified from insect cells (34) (Fig. 6 A and B). To test the motor activity of KIF27 toward microtubules specific to motile cilia, we purified tubulin heterodimers TUBA1A-TUBB4B from insect cells (60, 61) and polymerized them into microtubules with GMP-CPP, a slowly hydrolyzing analog of GTP (59, 62) (Fig. 6 B and C). Translocations of GMPCPP-stabilized TUBA1A-TUBB4B microtubules were observed using TIRF assays in the presence of KIF27_1-570GFP with 2 mM Mg-ATP, but not with 0 mM ATP or 2mM AMP-PNP, a nonhydrolyzable analog of ATP (SI Appendix, Fig. S11A). We further characterized the motor properties of KIF27_1-570GFP single particles using TIRF-microscopy assays (Fig. 6B). Analysis of kymographs revealed that in the presence of 1mM Mg-ATP, KIF27_1-570GFP exhibited processive translocations along GMP-CPP stabilized TUBA1A-TUBB4B microtubules, with a mean velocity of 25.5 ± 1.2 nm/s and a mean microtubule binding of 38.8 ± 2.7 s (Fig. 6 F–H). The in vitro velocity of KIF27 was orders of magnitude slower than KIF4 (63), indicating that KIF27 is unlikely to function as a ciliary transporter due to its slow-moving behavior.

Fig. 6.

KIF27 is an ATP-dependent kinesin motor. (A) Domain organizations of full-length human KIF27_1-1401 and the KIF27_1-570-GFP motor dimer used in the in vitro motility assays. (B) Schematic representation of in vitro TIRF set-up for the motility assay. GMPCPP-stabilized microtubules (gray) were attached to a PEGylated coverslip (blue) by Neutravidin links before addition of active KIF27 motor proteins (green). (C–E) Elution profiles from size exclusion chromatography and Coomassie blue-stained SDS-PAGE and immunoblot (IB) analyses of GFP-tagged KIF27_1-570 (peak volume, 57.86 mL) (C), biotinylated TUBA1A/TUBB4B (peak volume, 76.65 mL) and X-rhodamine conjugated TUBA1A/TUBB4B (peak volume, 76.94 mL) (D), GFP-tagged KIF27 (R206K)_1-570 (peak volume, 57.73 mL) (E). Arrows indicate the expected position of purified proteins. (F) (Top) Representative still image showing KIF27_1-570-GFP (cyan) binding to X-rhodamine-labeled microtubules (magenta) in the presence of 2 mM ATP. (Scale bar, 5 μm.) (Bottom) kymograph generated from the above KIF27 single molecule motility analysis. Horizontal Scale bar, 5 µm. Vertical Scale bar, 60 s. (G and H) Distribution of velocities (G) and binding durations (H) of KIF27_1-570 (n = 92 over 3 independent experiments). (I) (Top) Representative still image showing KIF27 (R206K)_1-570-GFP (cyan) binding to X-rhodamine-labeled microtubules (magenta) in the presence of 2 mM ATP. (Scale bar, 5 μm.) (Bottom) kymograph generated from the above KIF27 (R206K)_1-570-GFP single molecule motility analysis. Horizontal Scale bar, 5 µm. Vertical Scale bar, 60 s. (J and K) Distribution of velocities (J) and binding durations (K) of KIF27 (R206K)_1-570-GFP (n = 89 over three independent experiments). The average velocity or MT binding duration is shown as mean ± SE of mean.

Drosophila COS2 and mammalian KIF7 both act as microtubule associated scaffolds to mediate Hh signaling, raising the possibility that KIF27 may act as a scaffold in a motile cilia context. We next determined whether KIF27 ATP-dependent motility is required for its function in motile cilia. Two well-characterized mutations in the motor domain of the Drosophila kinesin-1 heavy chain Khc, E164A (64–66) and R210K (67, 68), can severely impair the ATP-dependent motility of the kinesin. E164 is located away from the ATP pocket of Khc and toward the motor–microtubule binding interface, and the E164A substitution results in a rigor kinesin that binds to ATP and microtubules with a higher affinity than wt. Khc R210 is within switch-I, a structural motif critical for ATP hydrolysis. Khc R210K homodimers lose the ability to hydrolyze ATP but remain associated with microtubules. Both E164 and R210 of Drosophila Khc are conserved in mammalian KIF27 (SI Appendix, Fig. S11B).

To test the effect of these substitutions on KIF27 ATP-dependent motor behaviors, we first expressed and purified the recombinant KIF27^E160A -GFP motor dimer analogous to the E164A Drosophila Khc mutant (SI Appendix, Fig. S11C). Using TIRF-microscopy assays, KIF27^E160A -GFP single particles showed a 36% reduction in motility associated with increased dwelling time on microtubules when compared to the control KIF27_1-570 motor dimers (SI Appendix, Fig. S11 D and E). In contrast, the KIF27^R206K-GFP, which contains the R206K substitution analogous to Drosophila R210K in Khc (Fig. 6E), completely blocked ATP-dependent KIF27 motility along TUBA1A-TUBB4B microtubules and resulted in a 40% decrease in microtubule binding lifetime (Fig. 6 I–K and SI Appendix, Fig. S9F). This Kif27^R206K allele can thus be used as a separation of function mutation to determine whether KIF27 motility is required for its role in motile cilia assembly.

KIF27 ATP-Dependent Motility Is Dispensable for Motile Cilia Assembly.

Based on the in vitro biochemical properties of KIF27^R206K motor dimers, we generated a mouse Kif27^R206K allele to investigate whether KIF27 motility is needed for motile cilia assembly in vivo (SI Appendix, Fig. S12A). In mouse tracheal epithelial cells (mTECs) derived from Kif27^GFP and Kif27^R206Khomozygous animals, KIF27^R206K-GFP protein was made and expressed at a comparable level as KIF27-GFP (SI Appendix, Fig. S12B). KIF27^R206K-GFP colocalized with CP110 at the distal centrioles at the deuterosome stage and with CEP164 at the TZ but was lost from the tip of nascent cilia (SI Appendix, Fig. S12 C–E). Using expansion microscopy analysis, we found that SPAG6 and CP microtubules were present in more than 66% of respiratory ciliated cells from the Kif27^R206K homozygous animals (Fig. 7 A and B and SI Appendix, Fig. S12F). Cilia with abnormal structures were present, but not to the extent observed in Kif27^−/− mice (Fig. 7A). Examination by tissue expansion microscopy demonstrated that in Kif27^R206K homozygous respiratory cilia, the domain organizations for RPGRIP1L and CEP290 at the TZ were largely restored (Fig. 7 C–E). The data suggest that ATP-dependent motility of KIF27 may contribute to cilia tip targeting of the motor protein but is dispensable for KIF27 function during cilia assembly.

Fig. 7.

KIF27 motility is dispensable for PCD-related cilia functions. (A) Expansion microscopy images of wt and Kif27^R206K/R206K nasal respiratory cilia labeled with acetylated α-tubulin (magenta) and SPAG6 (cyan). (Scale bar, 4 µm.) (B) Quantification of CP and SPAG6 organization in the wt and Kif27^R206K/R206K mice (n = 300 for each genotype, N = 3). (C) Expansion microscopy images of CEP290 and RPGRIP1L (cyan), acetylated α-tubulin (magenta), and CEP162 (yellow) from postnatal cilia with a cartoon for TZ organization on the right. (Scale bar, 2 µm.) (D and E) Average TZ height (D) and TZ height differences (E) for CEP290 and RPGRIP1L for wt and Kif27^R206K/R206K postnatal cilia. For (D and E), images shown in C were used for measurements n = 90 for each genotype, N = 3). Plots show 10 to 90% for (D and E) *P < 0.05, **P < 0.005, ***P < 0.0005, ****P < 0.0001, ns, not significant. (F) At P12, Kif27^−/− with hydrocephalus (asterisk) and littermate wt (Kif27 ^+/+). At 3 mo, Kif27^R206K/R206K homozygous mice and aged matched wt (Kif27^+/+). All Kif27^R206K/R206K mice were viable (n = 20). (G) H&E staining of brains and nasal cavities from 4-mo-old wt and Kif27^R206K/R206K mice. (Scale bar, 2 mm.) (H) SEM of nasal and ependymal cilia from 4-mo-old wt and Kif27^R206K/R206K mice. (Scale bar, 10 μm.) (I) Cilia beating frequencies (CBF) of postnatal wt (4.492 ± 2.046 Hz) and Kif27^−/− (3.347 ± 1.546 Hz) respiratory cilia. CBF in 4-mo-old wt (3.154 ± 1.834 Hz) and Kif27^R206K/R206K (2.661 ± 1.336 Hz) respiratory cilia. n = 60 for each genotype, N = 2; ***P < 0.0005, ns, not significant. CBF are shown as mean ± SD. Corresponding Movies S1–S4.

Unlike Kif27^−/−, all Kif27^R206K homozygous animals (N = 20) were viable with no hydrocephalus or nasal mucus accumulation (Fig. 7 F and G). Motile cilia from the nasal epithelium and brain ventricles of Kif27^R206K homozygous mice exhibited comparable morphologies and density as those observed for age-matched wt animals (Fig. 7H). Using wide-field live imaging, we evaluated cilia beating properties. We observed that beating frequencies were significantly reduced in Kif27^−/− respiratory tissues (Fig. 7I and Movies S1 and S2) but not in Kif27^R206K homozygous animals (Fig. 7I and SI Appendix, Fig. S12G and Movies S3 and S4). The enhanced cilia beating frequency and restored cilia structural integrity compared to Kif27^−/− mice may be sufficient to revert PCD-related pathology in Kif27^R206Khomozygous animals. To eliminate potential cilia damages introduced by mucus obstruction and tissue inflammation, mTECs from wt, Kif27^−/−, and Kif27^R206K were grown and differentiated for live-imaging analysis on ALI Day 9 (SI Appendix, Fig. S13A and Movies S7 and S9). Consistent with tissue-level observations, cilia beating frequencies (CBF) were comparable in wt and Kif27^R206KmTECs. Compared to both wt and Kif27^R206K, cilia beating frequency was significantly reduced in Kif27^−/− mTECs (SI Appendix, Fig. S13B and Movies S7 and S9). Using confocal microscopy, fluorescence intensities for CFAP20/acetylated α-tubulin signal was significantly reduced in Kif27^−/− mTEC but increased in Kif27^R206K mTEC when compared to the wt (SI Appendix, Fig. S13C). The data support a model in which a scaffold function of KIF27 at the motile cilia TZ is sufficient to drive cilia beating in vivo.

Discussion

Despite the vital functions motile cilia perform, how mammalian motile cilia are built is not understood. In this study, we demonstrate that mammalian KIF27 is required for the assembly of motile ciliary axoneme as well as the ciliary TZ. Kif27^−/− animals consistently exhibit hydrocephalus and nasal mucus accumulation, but do not show discernable laterality defects, indicating that the activities of KIF27 may be differentially regulated in a cell-type specific manner. In this regard, the heterogeneous requirement of KIF27 in mono and multiciliated cells parallels its paralogue KIF7, which exhibits tissue-specific regulations in Hh signaling (18, 41). Although KIF27 is an ATP-dependent motile kinesin, it does not use its motility for cilia assembly. Instead, our data suggest an assembly pathway in which a KIF27-mediated scaffold at the TZ is required to recruit structural proteins into mammalian motile cilia to drive motility. Since KIF27 homologues are present in many ciliated eukaryotes with motile cilia (69), KIF27 activities at the TZ may represent a conserved strategy to regulate motile cilia assembly in diverse species and cells. For species that display motile cilia but lack KIF27 (22, 70, 71), such as zebrafish, alterative pathways must be engaged to form the “9 + 2” axoneme with microtubule inner proteins.

Motile cilia of Kif27^−/− mice exhibit structure anomalies and defective waveforms, and gradually become shorter. In accordance with a synthetic KIF27 truncated protein (58), the in vitro velocity for KIF27 motor dimers was much slower than those reported for the Kinesin-4 motors or IFT Kinesin-II (72). Of the two homologues of KIF27, neither COS2 nor KIF7 is motile. Instead, both are required to tether the CI/GLI transcription factors to cytoplasmic and ciliary microtubules in Drosophila (73) and mammals (34), respectively, to mediate the graded response of Hh signaling. A parsimonious explanation is that, despite exhibiting ATP-dependent motility (34), a nonmotile scaffold function may be selected for and conserved in the KIF27/KIF7/COS2 kinesin lineage in cilia-dependent and independent contexts. We show that mice homozygous for the Kif27^R206K variant, which blocks ATP-dependent motility of KIF27, are alive without hydrocephalus or nasal mucus accumulation, suggesting that a nonmotile KIF27 is sufficient for motile cilia assembly in mammals. Nonetheless, a small fraction of abnormal axonemes is still present in Kif27^R206K homozygous respiratory cilia, which may be explained by a 40% decrease in in vitro microtubule binding lifetime observed for the KIF27^R206K motor dimers. In addition, KIF27^R206K failed to localize to the plus-ends of ciliary axonemes in vivo, suggesting that the ATP-dependent motility may account for the cilia tip tracking behavior of KIF27 pertinent to the maintenance and stabilization of ciliary axoneme. In vitro dynamic assays will clarify whether KIF27 can differentially regulate growth and stability of short and long microtubules, a combination of motor properties observed in some members of the Kinesin-4 family (35, 56–58, 63)^.

In motile ciliated cells, KIF27 localizes to the TZ in complex with CEP162, which crosslinks KIF27 with CEP290, a central scaffold of the MKS module at the TZ. CEP290 can bind to microtubules and membrane lipids through two separate regions (74, 75). Ablation of either CEP290 domain is sufficient to disrupt ciliogenesis and induce mislocalization of membrane-bound ciliary cargos, including opsin and rhodopsin of the photoreceptors (75). Our data suggest a model in which the KIF27–CEP162–CEP290 scaffold is formed in nascent cilia, and this KIF27-mediated scaffold may organize and stabilize the CEP290 complex at the TZ to facilitate cilia assembly and motility.

A connection between individual TZ proteins and cilia motility has been implicated in several species. Basalin, a Trypanosoma brucei specific TZ protein, is required for CP biogenesis by forming the basal plate, an electron-dense structure proposed to nucleate and stabilize CP microtubules (76, 77). NPHP1 and NPHP4 are required for cilia structure and axonemal organization in the connecting cilia of photoreceptor cells and sperm tail (78–80)^. Leber congenital amaurosis (LCA), a rare inherited form of early-onset eye diseases, causes severe visual impairment or blindness. A cohort of LCA patients with CEP290 variants were examined for the motile cilia morphologies based on their PCD-like symptoms, including chronic rhinitis and recurrent bronchitis. Nasal epithelial cells from these patients show axonemal defects, including missing CP, <9 microtubule doublets, as well as missing dynein arms (81). Consistent with these findings, axonemal defects were reported in brain ependymal motile cilia in a Cep290 knockout mouse model (82)^.

How does the TZ contribute to motile cilia assembly and function? Given that most TZ proteins have extensive coiled-coils to mediate protein–protein interactions, one scenario is that the TZ is a size-exclusion barrier that can selectively guide the motility machineries to specific loading docks for their incorporation into the growing axoneme, a function that has been proposed for the primary cilia TZ (11, 83). Consistent with this idea, TZ dysfunction may account for the depletion of selected ciliary cargos, such as MIPs, and ciliary retentions of IFT and BBsome subunits observed in the motile ciliome of Kif27^−/− mice and Cep290 Chlamydomonas mutants (84, 85). Examination of the assembly dynamics of CP apparatus and MIPs during ciliogenesis is needed to clarify whether they are bona fide IFT cargos. Heterologous KIF27 expressed in HEK293T cells can bind to STK36 (38), a predicted serine-threonine kinase implicated in rare cases of PCD (86). It remains an open question whether STK36 is tethered to the TZ through KIF27 to regulate phosphorylation-dependent barrier assembly and permeability, a phenomenon well characterized for other sorting machineries, such as the nuclear pore complex (87), axon initial segment (88), and septins (89, 90). An alternative and not mutually exclusive explanation is that KIF27, perhaps together with other TZ proteins, can directly interact with different sets of structural proteins to regulate their ciliary incorporation during ciliogenesis. Identification of the KIF27 interactome followed by functional validations will be required to shed light on the mechanistic details of KIF27-mediated TZ scaffold during motile cilia assembly.

Pathogenic variants have been identified for most genes encoding TZ proteins^, underscoring the crucial role of the ciliary TZ in embryonic development and tissue homeostasis^. However, the phenotypes and severity of TZ-related ciliopathies vary widely across TZ genes. The lack of a clear genotype–phenotype correlation suggests that the composition and function of the TZ are not uniform, but may exhibit heterogeneity across different cell types and organs. The unexpected connection between a KIF27-mediated TZ scaffold and the recruitment of motile cilia structural proteins highlights the gaps in our understanding of how TZ operates to regulate the assembly and function of mammalian motile cilia. Given the ubiquitous expression of TZ proteins in mammalian cells, it is likely that pathological variants of TZ genes can affect both primary and motile cilia and therefore, the manifestations of diseases associated with dysfunctional motile cilia may have been underestimated in ciliopathy patients. A mechanistic coupling between TZ dysfunction and motile ciliopathy represents an exciting area for future studies which may pave the way for expanding on the current list of motile ciliopathy genes and identifying new therapeutic strategies aimed at restoring cilia motility.

Materials and Methods

Mouse Strains.

All experimental procedures involving mice were performed in accordance with approved protocols by the Committee on the Use of Live Animals in Teaching and Research (CULATR) at the University of Hong Kong. See SI Appendix for detailed mouse strain information.

Cell Culture, Biochemistry, and Imaging Analyses.

Detailed methods for cell culture, protein purification, immunoblotting, immunoprecipitation, histology, scanning and transmission electron microscopy, TIRF microscopy, immunofluorescence confocal microscopy, lattice SIM superresolution microscopy, ultrastructure expansion microscopy, cilia live-imaging, mass spectrometry, cryotomography, and subtomogram averaging are included in SI Appendix.

Antibodies.

Primary and secondary antibodies used in immunoblots and immunofluorescence assays are summarized in Extended methods can be found in SI Appendix, Table S2.

Statistical Tests.

Methods for statistical analysis and numbers of samples measured in this study are specified in figure legends. The error bars indicate SD or SE. Two-tailed Student’s t tests were performed using Graphpad Prism software for data and are indicated in figure legends.

Supplementary Material

Appendix 01 (PDF)

Movie S1.

postnatal day-15 wt respiratory cilia beating.

Movie S2.

postnatal day-15 Kif27^-/- respiratory cilia beating.

Movie S3.

3-month old wt respiratory cilia beating.

Movie S4.

3-month old Kif27^R206K/R206K respiratory cilia beating.

Movie S5.

postnatal day-11 wt oviductal cilia beating.

Movie S6.

postnatal day-11 Kif27^-/- oviductal cilia beating.

Movie S7.

mTEC ALI Day 9 wt cilia beating.

Movie S8.

mTEC ALI Day 9 Kif27^R206K/R206K cilia beating.

Movie S9.

mTEC ALI Day 9 Kif27^-/- cilia beating.

Data, Materials, and Software Availability

EM data have been deposited in PDB (EMD-63946 and EMD-63944) (91, 92). Study data are included in the article and/or supporting information.