Mouse Cfap53 is specifically expressed in cells with motile cilia

We first examined the expression pattern of the Cfap53 gene in the mouse using a transgenic strain harboring Cfap53^lacZ bacterial artificial chromosome (BAC), in which lacZ has been knocked into the Cfap53 gene. In embryonic day (E) 8.0 embryos, Cfap53 was specifically expressed in the node, in particular within the pit cells that are located at the center of the node and have rotating monocilia on their apical surface (Fig 1A and 1B). In the adult mouse, we found lacZ expression in epithelial cells of the trachea, brain ventricles, the oviducts and in the testes (Fig 1C–1L), all of which are known to differentiate motile cilia. Thus, Cfap53 expression in the embryonic and adult mouse is strictly confined to cells with motile cilia.

Fig 1

Cfap53 is specifically expressed in cells with motile cilia.

Cfap53^lacZ mouse embryos and adult tissues were stained with the β-galactosidase substrate X-gal. The images include lateral (A) and ventral (B) views of E8.0 embryos; a sagittal section of the brain at 3 weeks of age (C), with higher magnification views of the boxed regions being presented in (D) and (E); the trachea at 8 weeks of age (F), with an image for a non-transgenic (Tg-) also being shown; sagittal section of the trachea at 8 weeks of age (G), with a higher magnification view shown in (H); the fallopian tube at 6 weeks of age (I); a sagittal section of the fallopian tube (J); and ducts of the testis at 10 weeks of age (K), with a higher magnification view shown in (L). Scale bars, 2 mm (K), 500 μm (C, G, J, L), 50 μm (D, E), 20 μm (H).

Differential requirement of CFAP53 for motility of 9+0 and 9+2 cilia

To uncover the function of CFAP53 in mouse development and physiology, we generated two types of mutants that lack either exon 2 or exons 2–8 of the Cfap53 gene (Fig 2A). Deletion of exon 2 alone is predicted to produce a severely truncated protein. Since Cfap53^Δex2/Δex2 and Cfap53^{Δex2-8/Δex2-8} mice exhibited indistinguishable phenotypes, the mutant allele lacking exon 2 is referred to as Cfap53^-/- hereafter. Cfap53^-/- mice were smaller at birth and showed laterality defects (Fig 2B and 2C). Half of Cfap53^-/- mice (7/14) exhibited situs inversus, while 15% (2/14) showed heterotaxy. In addition, all of the Cfap53^-/- mice examined (14/14) developed hydrocephalus around 3~4 weeks after birth and died by 6 weeks of postnatal life (Fig 2B and 2D).

Fig 2

Laterality defects, hydrocephalus, and ciliary motion defects in Cfap53 mutant mice.

(A) Genetic structure of the WT mouse Cfap53 locus and the generation of two types of knockout alleles lacking either exon 2 or exons 2 to 8. (B) Smaller body size and development of hydrocephalus in Cfap53^–/–mice at 4 weeks of age. (C) Laterality defects of Cfap53^–/–mice. (D) Survival curve for Cfap53^–/–mice (n = 14), with all animals dying by 6 weeks of age. (E) In situ hybridization analysis of Nodal expression in Cfap53^+/–and Cfap53^–/–mice at E8.0. Nodal expression was missing in the LPM of Cfap53^–/–embryos. Arrowheads indicate the node. (F) Immotility of node cilia in Cfap53^–/–embryos at E8.0. Red signals for the Cfap53^+/–embryo indicate motion trajectory in the corresponding movie. The red signal for the Cfap53^–/–embryo reflects the trajectory of a bubble. Dashed lines indicate the outline of the node. (G) Wave forms of tracheal and brain ventricle cilia determined from corresponding videos, S3 and S4 Videos. (H) Beat frequency for tracheal cilia of Cfap53^–/–and control mice (dots and triangles indicate Cfap53^+/+ and Cfap53^-/- mice, respectively) determined at 25°C. Data are presented as mean ± SD (n = 18 independent experiments); two tailed Student’s t-test (*p = 0.0041). See also S3 Video.

Laterality defects observed at newborn stages prompted us to examine several markers of L-R development at E8.0. Whereas Nodal is expressed bilaterally at the node and asymmetrically on the left lateral plate mesoderm (LPM) of wild-type (WT) embryos; in Cfap53^-/- embryos, however, Nodal expression at the node was maintained but that in the LPM was absent (3/3 embryos) (Fig 2E). Since these defects are likely to arise from impairment in the differentiation or motility of the node cilia, we examined these cilia as well as other kinds of motile cilia in the mutant mice. Node cilia are the only motile cilia found in WT embryos at E8.0, and they were completely immotile in Cfap53^-/- embryos (Fig 2F and S1 and S2 Videos). The length of node cilia in Cfap53^-/- embryos was slightly shorter than that in WT embryos (S1 Fig). Cilia within the trachea and those within the brain ventricles differentiated normally in the absence of CFAP53 and maintained the normal ciliary length (S1 Fig), but in contrast to the nodal cilia, retained motility, albeit with abnormal beating pattern. In particular, the beating angle of the tracheal cilia was smaller, making the effective stroke inefficient (Fig 2G and S3 and S4 Videos). In addition, the beating frequency of the mutant tracheal cilia was significantly reduced (Fig 2H). Cilia within the brain ventricles of the mutants also showed similar defects in their beating pattern, with the proximal portion of the ependymal cilia appearing rigid and with a reduced beating angle and inefficient stroke (Fig 2G), which is consistent with the observation that Cfap53^-/- mice develop hydrocephalus (Fig 2B). Cfap53^-/- males were infertile. Their sperm were morphologically abnormal lacking the tails (S2 Fig) and completely immotile (S5 and S6 Videos).

Localization of CFAP53 in node and tracheal cilia

We investigated the intracellular localization of CFAP53 in mouse tissues with an anti-CFAP53 antibody. In E8.0 embryos, CFAP53 protein was specifically detected at the node of Cfap53^+/+ embryos (Fig 3A–3C), but was absent from the node of Cfap53^-/- embryos (Fig 3D–3F). It was highly enriched at the base of the node cilia, most likely at the centriolar satellites [26], although much weaker signal could be detected along the axoneme (Fig 3G). Consistent with this observation, CFAP53 has been reported to be a novel centriolar satellites protein [25]. In fact, CFAP53-positive region at the base of node cilia merged with the centriolar satellites marked by PCM1 (Fig 3G–3I) and surrounded the basal body positive for γ-Tubulin (Fig 3J–3L).

Fig 3

Differential localization of CFAP53 in tracheal and node cilia.

(A–L) Immunofluorescence staining of the node of WT (A-C, G-L) or Cfap53^–/–(D-F) embryos at E8.0 with antibodies to CFAP53 (A, D, G, J), to acetylated (acet.) Tubulin (B and E), to PCM1 (H) and to γ-Tubulin (K). Scale bars, 10 μm (A-F). Higher magnification images are shown in insets. Scale bars, 5 μm (A-F). Note that CFAP53 is mainly localized at the base of nodal cilia (centriolar satellites positive for PCM1), and weakly in the axoneme. Scale bar, 2 μm (G-L). (M-R) Immunofluorescence analysis of an isolated tracheal MCC from a WT (M-O) or Cfap53^–/–(P-R) mouse. CFAP53 is detected at the base of cilia (the centriolar satellites) and in the axoneme. Scale bar, 5 μm.

In adult Cfap53^+/+ mice, CFAP53 protein was detected in MCCs of the trachea (Fig 3M–3O), but was absent from MCCs of Cfap53^-/- mice (Fig 3P–3R). CFAP53 was localized not only at the base of the tracheal cilia but also in the axonemes (Fig 3M–3O). Overall, CFAP53 localized mainly at the base (centriolar satellites) of node cilia, whereas it was also present prominently in the axonemes of tracheal cilia.

We also investigated intracellular localization of CFAP53 protein by establishing a transgenic strain harboring a Cfap53^Venus BAC, encoding functional CFAP53::Venus fusion protein (S3 Fig). Like the endogenous CFAP53 protein, the CFAP53::Venus fusion protein was expressed at the node, and was enriched at the base of node cilia although much weaker Venus signal was detected in the axoneme (S4 Fig). In the tracheal cilia of the transgenic mice, the fusion protein was detected at the base of MCC cilia and in their axonemes (S4 Fig). Unlike the endogenous CFAP53 protein, however, the fusion protein was absent in the distal end of the axonemes of tracheal cilia of Cfap53^Venus BAC, Cfap53^-/- mice (S4 Fig), suggesting that the CFAP53::Venus protein may fail to be transported efficiently to the distal region of tracheal cilia.

ODAs are lost from node cilia but mostly retained in tracheal cilia of Cfap53^-/- mice

Given that node cilia lose motility in the absence of CFAP53, we next examined the presence or absence of DNAH5, a major axonemal dynein heavy chain protein, by immunostaining. In WT node cilia, DNAH5 is localized throughout the axoneme (Fig 4A–4C). In Cfap53 mutant node cilia, however, DNAH5 was not detected in the axoneme (Fig 4D–4F). In tracheal cilia, on the other hand, DNAH5 was detected throughout the axoneme, in both control (Fig 4G–4I) and in Cfap53^-/- mice (Fig 4J–4L). Since DNAH5 is essential for ciliary motility [27,28], the absence or presence of DNAH5 in the two types of cilia correlated with effects of the loss of CFAP53 on their ability to move.

Fig 4

Localization of DNAH5 in ciliated cells of the trachea and node.

(A-F) Immunofluorescence staining with antibodies to DNAH5 (green) and to acetylated (acet.) Tubulin (magenta) of the node at E8.0 and adult trachea (G-L) from Cfap53^–/–and control mice. (D-F, J-L) DNAH5 was absent from node cilia but maintained in tracheal cilia of Cfap53^–/–mice. Scale bars, 20 μm. Insets show higher magnification views. Scale bars, 2 μm.

We also examined the localization of two other important ODA dynein heavy chains, DNAH11 and DNAH9. Since antibodies suitable for immunostaining of DNAH11 in mouse embryos are not available, we generated the Dnah11^Venus allele, which encodes a functional DNAH11::Venus fusion protein (S5 Fig), and examined its localization. DNAH11 localized along the entire axoneme of node cilia (Fig 5A–5C) as previously described [11,29], while it localized only to the proximal portion of trachea cilia (Fig 5G–5I), as reported earlier from studies with human respiratory MCCs [11]. However, DNAH11 was absent from node cilia of Cfap53^-/- embryos (Fig 5D–5F). By contrast, proximal localization of DNAH11 in tracheal cilia was maintained in Cfap53^-/- mice (Fig 5J–5L).

Fig 5

Localization of DNAH11 in ciliated cells of the trachea and node.

(A-F) Immunofluorescence staining with antibodies to GFP (green) and to acetylated (acet.) Tubulin (magenta) for the node at E8.0 and adult trachea (G-O) of Cfap53^–/–; Dnah11^Venus and control mice. (D-F, J-L) DNAH11::Venus was absent from node cilia but was maintained in tracheal cilia of Cfap53^–/–mice. In (C) and (F), the arrowhead indicates a node cilium. Scale bars, 10 μm (A-F) or 5 μm (G-O). Insets represent higher magnification views.

In humans, DNAH9 is localized to the distal region of tracheal cilia, and in PCD patients with DNAH9 mutations, the motility of the tracheal cilia is affected. Many of these patients also exhibit situs defects implying a requirement for the DNAH9 protein in human nodal cilia motility [30]. In tracheal cilia of Cfap53^-/- mice, distal localization of DNAH9 was maintained, although its level was reduced compared with that apparent in control mice (S6 Fig). Unexpectedly, we failed to detect DNAH9 protein in node cilia, either with a DNAH9-specific antibody (S6 Fig) or by live imaging of a DNAH9::Venus fusion protein encoded by a Dnah9^Venus transgene (S7 Fig). To confirm the absence of DNAH9 at the node of the mouse embryo, we generated a mutant mouse deficient in this protein (Dnah9 KO mouse; Fig 6A). Dnah9 KO mouse showed normal laterality (9/9 newborn mice; Fig 6B–6D), unlike in humans where DNAH9 mutations result in laterality defects [30]. When Dnah9 KO mice were dissected, L-R asymmetries of the lung, the heart apex, great arteries, azygous vein, stomach, liver and spleen were found to be normal (Fig 6D). Around one month after birth, four of the six Dnah9 KO mice examined showed hydrocephalus (Fig 6E). Tracheal and ependymal cilia from Dnah9 KO mice were motile, but exhibited abnormal beat pattern with reduced beating angle and insufficient effective stroke (Fig 6F, S7–S10 Videos). The beat frequency of mutant tracheal cilia was reduced (Fig 6G). Sperm from a Dnah9 KO male was morphologically normal and retained normal motility (S11 and S12 Videos), which is consistent with the previous report that DNAH9 is not expressed in mouse sperm [31].

Fig 6

Phenotype of Dnah9 mutant mouse.

(A) The exon-intron structure of mouse Dnah9 gene and generation of knockout allele lacking exon 30. Note that the motor domains of DNAH9 are encoded by exons 29–32 and exons 34–35. (B) Eight pups generated by CRISPR-mediated mutagenesis were genotyped by PCR. (C) External appearance of the same eight pups whose genotype is shown in (B). Arrowheads denote the position of the stomach. Note that all the Dnah9^-/- (KO) mice have the stomach at the normal position. (D) Laterality of visceral organs in WT and Dnah9^-/- mice. The arrowheads indicate the position of the heart apex. (E) Dnah9^–/–mice developed hydrocephalus (arrowhead). (F) Wave forms of tracheal and brain ventricle cilia of Dnah9^+/+ and Dnah9^–/–mice. See also S7–S10 Videos. (G) Beat frequency for tracheal cilia of Dnah9^–/–and control mice. Data are presented as mean ± SD (n = 14 independent experiments); two tailed Student’s t-test (*p<0.05). See also S7 and S8 Videos.

We next studied the ultrastructure of node and tracheal cilia of WT and Cfap53 mutants using transmission electron microscopy (TEM). The axonemes of node cilia in WT embryos have a 9+0 configuration with ODAs attached to the nine peripheral microtubule doublets (Fig 7A and 7B). However, in node cilia from Cfap53^-/- embryos, ODAs were missing from most of the sections examined (Fig 7C and 7D and 7J). Tracheal cilia from WT mice show 9+2 configuration with the CPA and ODAs attached to the peripheral microtubule doublets (Fig 7E and 7F). In tracheal cilia of Cfap53^-/- mice, most ODAs were retained (Fig 7G, 7I and 7K), consistent with our observations that these cilia of the mutants are able to move, albeit with an altered motility pattern (Fig 2G). When tracheal cilia were examined along the proximal-distal direction, there was no significant preference in the loss of ODAs (S8 Fig). Thus, the partial loss of ODAs from tracheal cilia (with most axonemes having five to eight ODAs) allowed them to maintain their motility, but impaired their motion pattern.

Fig 7

ODAs are completely lost in node cilia but largely maintained in tracheal cilia of Cfap53^–/–mice.

(A–D) TEM of node cilia of Cfap53^+/+ (A) and Cfap53^–/–(C) embryos at E8.0. Higher magnification views of corresponding doublet microtubules indicated by the arrowheads are shown in (B) and (D), respectively, together with schematic diagrams of the microtubule pairs and ODAs (blue protrusions). Scale bars, 100 nm. (E–I) TEM of tracheal cilia of adult Cfap53^+/+ (E) and Cfap53^–/–(G) mice. Higher magnification views of corresponding doublet microtubules indicated by the arrowheads are shown in (F), (H), and (I) together with schematic diagrams of the microtubule pairs and ODAs. Scale bars, 100 nm. (J and K) Distribution of the number of ODAs per axoneme for node (J) and tracheal (K) cilia of Cfap53^–/–and Cfap53^+/+ mice determined from TEM images. (L) Nine peripheral doublet microtubules were numbered from DMT1-DMT9, on the base of their relative position to the central pair: DMT1 is the doublet microtubule located at the position perpendicular to the central pair microtubules. The frequency of ODA loss at each DMT of Cfap53^–/–tracheal cilia is shown by a circular plot. Note that DMTs with the line indicated by asterisk (#3 and #8) are preferentially lost (ρ = 0.04263 by Rayleigh’s test). (M) Distribution of the number of ODAs per axoneme for tracheal cilia of Cfap53; Rsph4a double mutant mice determined from TEM images.

Interestingly, ODAs at the position 3 and 8 were preferentially lost in tracheal cilia of Cfap53^-/- mice (Fig 7L). It may be that ODAs are weakly attached to microtubules in the absence of CFAP53, and ODAs at position 3 and 8 are preferentially detached from microtubules because planar beating likely takes place along the ODA 3–8 axis. To test this, we examined the tracheal cilia of Cfap53^-/-, Rsph4a^-/- double mutant (DKO) mice, since tracheal cilia exhibit rotational movement instead of planar beating in the absence of RSPH4a [17]. ODAs in tracheal cilia of Cfap53^-/-; Rsph4a^-/- DKO mouse were more severely lost than those of Cfap53^-/- mouse (Fig 7M).

CFAP53 associates with axonemal microtubules of tracheal cilia and interacts with dynein proteins and the docking complex (DC)

Given that CFAP53 is an axoneme-associated protein in tracheal cilia, we investigated its localization within these cilia more precisely. First, we examined whether CFAP53 interacts with microtubules. Isolated tracheal cilia were treated with high salt (0.6 M NaCl or NaI), and then subjected to centrifugation to obtain soluble and insoluble fractions. Immunoblot analysis revealed that, after treatment with 0.6M NaCl, microtubules remained insoluble, with tubulin being detected in the precipitate fraction (Fig 8A). However, the dynein complex became detached from the microtubules, with ODA dynein intermediate chain 2 (DNAIC2) being detected in the soluble fraction. Under this condition, CFAP53 was present in the precipitate, suggesting that CFAP53 protein remained associated with microtubules. When the tracheal cilia were treated with 0.6M NaI, which disrupts most protein-protein interactions, microtubules were dissociated and tubulin was detected in the soluble fraction. Under this condition, DNAIC2 and CFAP53 were also found largely in the soluble fraction (Fig 8A). These results suggest that CFAP53 is tightly associated with microtubules in tracheal cilia.

Fig 8

Association of CFAP53 with axonemal microtubules of tracheal cilia as well as with DNAIC proteins and TTC25.

(A) Isolated axonemes of tracheal cilia were treated (or not, Cont) with 0.6 M NaCl or NaI and then centrifuged at 15000 × g for 12 min at 4°C, and the resulting precipitate and supernatant fractions were subjected to immunoblot analysis with antibodies to CFAP53, to DNAIC2 (ODA marker) and to α-Tubulin. (B–E) HEK293T cells transfected with expression vectors for HA-tagged CFAP53 or MYC epitope-tagged DNAIC1, DNAIC2, DYNLL1 (cytoplasmic dynein light chain), TTC25, CCDC114, or CCDC151 (or with the corresponding empty vectors), or GFP-tagged DNAH11 N-terminus fragments 1–1000 aa or 1001–1700 aa, as indicated, were subjected to immunoprecipitation (IP) with antibodies to HA or to MYC or to GFP, and the resulting precipitates as well as the original whole cell lysates (WCLs) were subjected to immunoblot analysis with antibodies to HA, MYC or GFP. CFAP53 interacted specifically with the axonemal dyneins DNAIC1, DNAIC2 and DNAH11 N-terminus and the DC member TTC25.

The tight association of CFAP53 protein with microtubules suggests that CFAP53 may interact with ODA dyneins and/or DC component(s). We therefore examined potential physical interaction of CFAP53 with three ODA dynein components (the dynein intermediate chains DNAIC1, DNAIC2 and the heavy chain DNAH11) and three components of the DC (TTC25, CCDC114 and CCDC151) by co-immunoprecipitation analysis of transfected HEK293T cells (Figs 8B–8E, S9). TTC25 is thought to be the DC component most proximal to the axonemal microtubules [32], while CCDC151 is the most distal DC component [13]. Over-expressed CFAP53 interacted with over-expressed DNAIC1, DNAIC2, DNAH11 and TTC25, but not with CCDC114 or CCDC151 (Fig 8B–8E, S9). These results suggest that in the axoneme of tracheal cilia, CFAP53 is embedded in microtubules and may directly interact with the DC via TTC25. CFAP53 at the centriolar satellites of node and tracheal cilia may also interact with TTC25 and ODA proteins, facilitating their transport into the axoneme.

Differential requirement of CFAP53 for axonemal localization of TTC25 in the node and tracheal cilia

Given that CFAP53 physically interacts with TTC25, we next examined the genetic relationship between CFAP53 and TTC25. In Ttc25^-/- mutant mice [32], localization of CFAP53 was maintained in node cilia (Fig 9A–9F), suggesting that the correct localization of CFAP53 does not require TTC25. In contrast, the subcellular localization of TTC25 was dependent on CFAP53, as revealed by the observation that in the node cilia of Cfap53^-/- embryos, TTC25 protein was absent from the axonemes (Fig 9G–9L). Given that TTC25 is a DC component located most proximally to the axonemal microtubules and mutation of TTC25 in mice and humans inhibits the association of ODAs with axonemes [32], the lack of TTC25 from the axonemes of Cfap53^-/- node cilia is likely to be responsible for their severe lack of ODAs. On the other hand, in tracheal cilia, axonemal localization of TTC25 protein does not strictly require CFAP53 function, and TTC25 was found in the axonemes of tracheal cilia of Cfap53 mutants, although at reduced levels (S10 Fig). This reduced level of TTC25 likely accounts for the partial loss of ODAs from the tracheal cilia (Fig 7K).

Fig 9

Genetic interaction between Cfap53 and Ttc25 in node cilia.

(A–F) Immunofluorescence staining with antibodies to CFAP53 (green) and to acetylated (acet.) Tubulin (magenta) of the node from Ttc25^–/–and control embryos at E8.0. CFAP53 was localized at the axoneme and base of node cilia of Ttc25^–/–embryos. (G–L) Immunofluorescence staining with antibodies to TTC25 (green) and to acetylated (acet.) Tubulin (magenta) of the node from Cfap53^–/–and control embryos at E8.0. TTC25 was lost from the axoneme of node cilia in the Cfap53^-/-. Scale bars, 10 μm. Insets show higher magnification views. Scale bars, 2 μm.

Ethics statement

All mouse experiments were approved by the Institutional Animal Care and Use Committee (Permission number: A2016-01-6) and carried out in accordance with guidelines of RIKEN BDR.

Mice

Mice were maintained at the Animal Facility of Riken Center for Biosystems Dynamics Research (BDR), Japan, under a 12-h light, 12-h dark cycle and were provided with food and water ad libitum. Two types of Cfap53 knockout mice, lacking exon 2 or exons 2 to 8, were generated with the CRISPR-Cas9 system as performed with guide RNA 1 (located upstream of exon 2 5’-CCTCTTAATTTTTACTTATTGTA-3’), guide RNA 2 (5’-GCTTCAACTCCTGCACCACCAGG-3’) and guide RNA 3 (5’-GGTGAGCCAAAATATGGGCC-3’). For generation of Cfap53^Venus BAC transgene, the Venus coding sequence was inserted immediately before the stop codon of Cfap53 in the BAC clone RP23-99F21. For generation of the Cfap53^lacZ transgene, an IRES (internal ribosome entry site)-lacZ cassette was inserted immediately after the stop codon of Cfap53 in the same BAC clone. The Dnah11^Venus allele was generated by inserting Venus into exon 43 of Dnah11 gene in frame. The Dnah9^Venus allele was generated by inserting Venus at the carboxyl terminus of DNAH9. The construction of the Ttc25^-/- mice has been described previously [32]. Dnah9 knockout mice, lacking exon 30, were generated with the CRISPR-Cas9 system as performed with guide RNA 1 (located upstream of exon 30 5’-TCAGACATAGAAGTGCGAGATGG-3’) and guide RNA 2 (5’-GATCGATCTTACAGCTTGTCTGG-3’). Cfap53 mutant mice were generated and maintained on a BC57BL background, while Cfap53::Venus Bac transgenics were generated by injecting BAC DNA into the pronuclei of (C57BL X C3H) F1 mouse eggs.

In situ hybridization and staining for lacZ expression

Whole-mount in situ hybridization was performed according to standard protocols with specific RNA probes for Nodal. Staining for lacZ expression was performed with the β-galactosidase substrate X-gal (5-bromo-4-chloro-3-indolyl-β-D-galactopyranoside) as described previously [40]. Serial sections (thickness of 10 μm) of the brain, trachea, oviduct, and testis were prepared after X-gal staining.

Antibodies

cDNA for the C-terminal region of mouse CFAP53 was cloned into the bacterial expression vector pMaL-c2X (New England Biolabs) with the use of the forward and reverse primers 5'-AGAGTCTATGCAGCGGGAGT-3' and 5'-TAGAGGAGGCTTGTCAGGGTA-3', respectively. CFAP53 expressed as a fusion protein with maltose binding protein in Escherichia coli was purified from the bacterial cells with the use of amylose resin (New England Biolabs) and injected into rabbits for the generation of antibodies. Specific antibodies were purified from rabbit serum by affinity chromatography. Additional antibodies used in this study are rabbit anti-GFP (Invitrogen, A-11122), mouse anti-acetylated Tubulin (Sigma, T6793), mouse anti- γ-Tubulin (Sigma, T5326), mouse anti-PCM1 (Santa Cruz, SC-398365), mouse anti-human DNAI2 (Abnova, H00064446-MO1), mouse anti-α-Tubulin (Sigma, T5168), mouse anti-HA (Santa Cruz, SC7392), mouse anti-GFP (Abcam, ab1218), rabbit anti-GFP (Santa Cruz SC-8334) mouse anti-MYC (Santa Cruz, SC-40), rabbit anti-HA (Santa Cruz, SC-805) and rabbit anti-MYC (Santa Cruz, SC-289). Antibodies to DNAH5 and DNAH9 were kindly provided by H. Takeda, described previously [41]. Rabbit antibodies to TTC25 were described previously [42]. Alexa Fluor-conjugated secondary antibodies for immunofluorescence staining were obtained from Invitrogen.

Immunofluorescence staining

E8.0 embryos were collected in ice-cold phosphate buffered saline (PBS) and fixed in 4% paraformaldehyde for 30 min on ice. They were then dehydrated at -20°C in methanol, exposed to a blocking reagent (Perkin Elmer FP1020), for 1 h at 4°C, incubated overnight at 4°C with primary antibodies diluted in blocking reagent (Perkin Elmer FP1020), washed for 6 h with PBS containing 0.1% Triton X-100 (PBST), incubated overnight at 4°C with Alexa Fluor–conjugated secondary antibodies diluted in blocking reagent (Perkin Elmer FP1020), solution or Can Get Solution TM A/B (Toyobo), and washed again. The nodes of the stained embryos were excised and placed on glass slides fitted with silicone rubber spacers, and covered with cover glass.

Tracheae of adult mice were isolated, cut longitudinally in ice-cold PBS, fixed in 4% paraformaldehyde for 1 h at 4°C, permeabilized with 1% PBST for 10 min at 4°C, and then processed for immunostaining as for the node. The tissue was then placed on a glass slide with silicone rubber spacers and covered with a cover glass. For preparation of cryosections of the immunostained trachea, samples were exposed overnight at 4°C to sucrose at increasing concentrations of up to 30%, embedded in OCT compound, frozen in liquid nitrogen_, and sectioned at a thickness of 10 μm. The sections were dried, washed with PBS, mounted in Antifade Prolong Gold (Invitrogen), and covered with a cover glass.

Tracheal epithelial cells, isolated as described previously [43], were fixed in methanol for 10 min at –20°C, exposed to 1% bovine serum albumin at room temperature for 30 min, and incubated for 1 h at room temperature with primary antibodies diluted in Can Get Solution TM A (Toyobo). After washing with PBST, the cells were incubated for 1 h at room temperature with Alexa Fluor-conjugated secondary antibodies diluted in Can Get Solution A and then washed again. All immunostained samples were observed with Olympus FV1000, FV3000 or Zeiss LSM 880 with Airyscan microscopic systems.

Imaging of ciliary motion

Trachea and brain were collected into HEPES-buffered Dulbecco’s modified Eagle’s medium (DMEM) or high-glucose DMEM (Nacalai), respectively. Embryos were collected into DMEM-HEPES with 10% Fetal bovine serum (FBS). Motility of cilia was examined at 25°C with a high-speed CMOS camera (HAS-500M or HAS-U2, DITECT) at a frame rate of 100 frames/s for node cilia or 500 frames/s for tracheal and brain ependymal cilia. The pattern of ciliary motion was traced and analyzed with ImageJ and Photoshop CC (Adobe).

TEM

E8.0 embryos and adult trachea were fixed with 2% paraformaldehyde and 2.5% glutaraldehyde (and 0.1% tannic acid for the trachea) in 0.1 M cacodylate buffer. The samples were then washed with 4% sucrose in 0.1 M cacodylate buffer, exposed to 1% osmium tetroxide, stained en bloc with 0.5% uranyl acetate, dehydrated with a graded series of ethanol, and embedded with the use of a Poly/Bed 812 Embedding Kit (Polysciences). Ultrathin (60 nm) sections were prepared with a diamond knife, mounted on 200-mesh copper grids, stained with 2% uranyl acetate and 3% lead citrate, and observed with a JEM-1400 plus microscope (JEOL).

Immunoblot analysis

Ciliated cell cortices were prepared from the mouse trachea as described previously [44], with some modifications. In brief, cilia were removed by vortex mixing in a solution containing 20 mM Hepes-NaOH (pH 7.5), 10 mM CaCl₂, 1 mM EDTA, 50 mM NaCl, 1 mM dithiothreitol and cOmplete protease inhibitor cocktail (Roche), and the isolated cilia were resuspended in a solution containing 20 mM Hepes-NaOH (pH 7.6), 50 mM KCl, 4 mM MgSO₄, 0.5 mM EDTA, 1 mM dithiothreitol, and cOmplete (Roche). The membrane of the cilia was removed by exposure to 0.1% Triton X-100 for 10 min on ice, and the axonemes were then washed with resuspension buffer and treated for 10 min on ice with resuspension buffer supplemented with high salt (0.6 M NaCl or NaI). The high salt-treated axonemes were subjected to centrifugation at 15,000 x g for 12 min at 4°C, and the resulting pellet and supernatant were subjected to immunoblot analysis with antibodies to CFAP53, to DNAIC2, and to α-Tubulin. Immune complexes were detected with horseradish peroxidase-conjugated antibodies to rabbit (NA934, GE HealthCare) or mouse (NA931, GE HealthCare) immunoglobulin G.

Co-immunoprecipitation analysis

The coding sequence of mouse Cfap53 was cloned into the pxj40-HA vector (Roche), whereas those of mouse Dnaic1, Dnaic2, Dynll1, Ccdc114, Ccdc151 and Ttc25 were cloned into pxj40-MYC (Roche) and Dnah11(N-terminus 1-1000aa), Dnah11(N-terminus 1001-1700aa) were cloned into pxj40-GFP (Roche). The indicated combinations of plasmids were introduced into HEK293T cells in 10-cm dishes by transfection for 24 h with the use of the Lipofectamine 2000 reagent (Thermo Fisher Scientific). The cells were then lysed in 1 ml of RIPA buffer (Thermo Fisher Scientific) supplemented with cOmplete Mini protease inhibitors (EDTA-free, Roche) and subjected to immunoprecipitation with 2 μg of mouse monoclonal antibodies to HA, MYC or GFP together with 30 μl of protein A-agarose beads (Roche). The resulting immunoprecipitates as well as the cell lysates were then subjected to immunoblot analysis with rabbit polyclonal antibodies to HA, MYC or GFP. Immune complexes were detected with horseradish peroxidase-conjugated secondary antibodies to mouse (Promega, W4028) or rabbit (Promega, W4018) immunoglobulin G.