ABSTRACT
Vertebrate motile cilia are classified as (9+2) or (9+0), based on the presence or absence of the central pair apparatus, respectively. Cryogenic electron microscopy analyses of (9+2) cilia have uncovered an elaborate axonemal protein composition. The extent to which these features are conserved in (9+0) cilia remains unclear. CFAP53, a key axonemal filamentous microtubule inner protein (fMIP) and a centriolar satellites component, is essential for motility of (9+0), but not (9+2) cilia. Here, we show that in (9+2) cilia, CFAP53 functions redundantly with a paralogous fMIP, MNS1. MNS1 localises to ciliary axonemes, and combined loss of both proteins in zebrafish and mice caused severe outer dynein arm loss from (9+2) cilia, significantly affecting their motility. Using immunoprecipitation, we demonstrate that, whereas MNS1 can associate with itself and CFAP53, CFAP53 is unable to self-associate. We also show that additional axonemal dynein-interacting proteins, two outer dynein arm docking (ODAD) complex members, show differential localisation between types of motile cilia. Together, our findings clarify how paralogous fMIPs, CFAP53 and MNS1, function in regulating (9+2) versus (9+0) cilia motility, and further emphasise extensive structural diversity among these organelles.
INTRODUCTION
Motile cilia are biological nanomachines, capable of rhythmic beating to generate fluid circulation over epithelia or facilitate locomotion of individual cells and organisms. For example, motile cilia lining the airways and brain ventricles clear mucus and circulate cerebrospinal fluid, respectively, whereas those on protozoans, invertebrate larvae and sperm are required for propulsion through fluid medium (Zhou and Roy, 2015). In humans, defective motile cilia are associated with diverse pathological consequences that include respiratory disease, hydrocephalus and infertility, collectively called primary ciliary dyskinesia (Wallmeier et al., 2020). The core or axoneme of a prototypical motile cilium comprises a circular arrangement of nine peripheral microtubule doublets (PMDs) (complete A tubule associated with an incomplete B tubule) that surround a pair of central singlet microtubules – the classic (9+2) arrangement (Fig. 1A). Inner dynein arms (IDAs) and outer dynein arms (ODAs), anchored to the PMDs, are responsible for motility of the organelle. One variation on the (9+2) configuration is the (9+0) arrangement, which lacks the central pair and its associated structures, such as radial spokes (Fig. 1B). Loss of the central pair is believed to allow rotational movement of (9+0) cilia, in contrast to the whiplash-like, back-and-forth planar beating of (9+2) cilia. This idea emerged from the observation that central pair loss from otherwise (9+2) cilia confers on them a rotary beat pattern (Shinohara et al., 2015). The best characterised (9+0) cilia are found within the left-right (L-R) organiser (LRO) in embryos of several vertebrate species [Kupffer's vesicle (KV) of teleost fishes and the ventral node of mouse embryos], where their rotary motion drives a vectorial flow of extra-embryonic fluid for initiating asymmetric development of visceral organs (Zhou and Roy, 2015) (for a summary of (9+0) and (9+2) cilia distribution in zebrafish and mouse, see Table 1). Consequently, people with primary ciliary dyskinesia often have mirror image reversal (situs inversus) or randomisation (situs ambiguus or heterotaxy) of visceral organ disposition, presumably as a result of LRO cilia dysfunction (Wallmeier et al., 2020).
Motile cilia structure, CFAP53 and MNS1 protein disposition within the A tubule and their paralogous nature. (A,B) Diagram of (9+2) (A) and (9+0) motile cilia (B). The peripheral microtubule doublet (PMD), central pair (CP), inner dynein arm (ID), outer dynein arm (OD) and radial spoke (RS) are indicated. (C) Diagrammatic transverse section through the PMD, showing MNS1 and CFAP53 localisation within the A tubule. (D) Diagrammatic longitudinal section through the A tubule (microtubule protofilaments A6-A8), illustrating CFAP53 and MNS1 arrangement. C, C terminus; N, N terminus. (E) MSA of fMIP family members bearing the LRQ motif. Asterisks indicate invariably conserved residues. Proteins are designated by their UniProt identifiers. Colouring schemes are as per ClustalX parameters. Species represented are Homo sapiens, Mus musculus, Xenopus tropicalis, Danio rerio, Tetrahymena thermophilia and Chlamydomonas reinhardtii.
Motile cilia structure, CFAP53 and MNS1 protein disposition within the A tubule and their paralogous nature. (A,B) Diagram of (9+2) (A) and (9+0) motile cilia (B). The peripheral microtubule doublet (PMD), central pair (CP), inner dynein arm (ID), outer dynein arm (OD) and radial spoke (RS) are indicated. (C) Diagrammatic transverse section through the PMD, showing MNS1 and CFAP53 localisation within the A tubule. (D) Diagrammatic longitudinal section through the A tubule (microtubule protofilaments A6-A8), illustrating CFAP53 and MNS1 arrangement. C, C terminus; N, N terminus. (E) MSA of fMIP family members bearing the LRQ motif. Asterisks indicate invariably conserved residues. Proteins are designated by their UniProt identifiers. Colouring schemes are as per ClustalX parameters. Species represented are Homo sapiens, Mus musculus, Xenopus tropicalis, Danio rerio, Tetrahymena thermophilia and Chlamydomonas reinhardtii.
Cryogenic electron microscopy (cryo-EM) has revolutionised our understanding of ciliary structural organisation. Studies of motile ciliary axonemes from various sources, such as the green alga Chlamydomonas, bovine and human airways as well as sea urchin and mammalian sperm, have provided us with a detailed molecular map of the different proteins and protein complexes that make up the motility apparatus (such as dynein arms, radial spokes, and the nexin-dynein regulatory complex). These studies have also identified many proteins that decorate the outer and inner surfaces of the axonemal microtubules (Gui et al., 2022, 2021; Leung et al., 2023; Ma et al., 2019; Walton et al., 2023; Zhou et al., 2023). These microtubule-associated proteins are thought to facilitate anchoring of important motility proteins, such as the dynein arm complexes, as well as provide structural strength so that the microtubules can withstand motility-induced deformation. These data underscore an evolutionarily conserved axonemal architecture, but they have also revealed notable differences between species (for example, between Chlamydomonas and mammals) and also between cilia from different regions of the same organism (for example, respiratory cilia versus sperm flagella from mammals) (Gui et al., 2021; Leung et al., 2023; Ma et al., 2019; Walton et al., 2023; Zhou et al., 2023). However, that all of this information pertains exclusively to (9+2) cilia as they are amenable to large-scale isolation in pure form, such as through deciliation of algal cells and sperm or trachea from slaughtered mammals. By contrast, (9+0) cilia are relatively limited in numbers and located in tissues and organs from which their isolation is difficult (in the mouse embryo LRO, they number between 200 and 300; Shinohara et al., 2012), making ultrastructural characterisation challenging. This leaves open the question of whether the structural organisation of ciliary proteins varies between (9+0) and (9+2) axonemes, and, if so, to what degree.
Functional studies are indeed indicating that the localisation and motility requirements for even highly evolutionarily conserved components can be rather different between the two cilia types. For instance, CFAP53 is a major A tubule filamentous microtubule inner protein (fMIP) (positioned in the groove between protofilaments A6 and A7) of cilia from protozoans to mammals, as revealed by cryo-EM (Gui et al., 2021; Ma et al., 2019) (Fig. 1C,D). Consistent with this, immunolocalisation of the endogenous protein and epitope-tagged functional variants has shown that it localises along axonemes of (9+2) cilia in zebrafish, mice and humans (Ide et al., 2020; Narasimhan et al., 2015). Intriguingly, CFAP53 is also a centriolar satellites protein, associated with basal bodies of (9+2) cilia, and in (9+0) cilia it was found to localise exclusively (zebrafish KV cilia) or preferentially (mouse nodal cilia) to this structure (Ide et al., 2020; Narasimhan et al., 2015; Silva et al., 2016). Furthermore, loss-of-function studies with zebrafish and mice have shown that, whereas (9+0) LRO cilia lose their motility, (9+2) cilia remain motile, albeit with altered beat frequency and amplitude (Ide et al., 2020; Narasimhan et al., 2015; Noel et al., 2016). These defects ensue from a nearly complete [from (9+0) cilia)] or partial [from (9+2) cilia] ODA loss, consistent with biochemical and cryo-EM data implicating association of CFAP53 with the outer dynein arm docking (ODAD) member ODAD4, and also the ODA dyneins themselves (directly or through linker MIPs) (Gui et al., 2021; Ide et al., 2020; Narasimhan et al., 2015). In line with this, people with CFAP53 mutations exhibit strong laterality defects, signifying disruption of LRO cilia motility, but have a mild alteration in their respiratory (9+2) cilia motility (Narasimhan et al., 2015; Noel et al., 2016; Perles et al., 2012).
We were intrigued by the differential localisation and function of CFAP53 in the two cilia types and set out to investigate why it is largely dispensable in (9+2) cilia. Here, we show that CFAP53 is paralogous to another evolutionarily conserved A tubule fMIP, MNS1, that binds the groove between protofilaments A7 and A8, immediately adjacent to A6 and A7 (occupied by CFAP53) (Gui et al., 2021; Ma et al., 2019) (Fig. 1C,D). Unlike CFAP53, which is also a centriolar satellites protein, MNS1 localises mostly to axonemes of both cilia types in zebrafish and mice. However, like CFAP53, MNS1 loss significantly impacted LRO cilia motility, but (9+2) cilia were much less severely affected. Consistent with redundancy, deficiency of both proteins from (9+2) cilia caused complete loss of ODAs in mice and a range of motility defects, from highly abnormal beating to total paralysis, in zebrafish. Finally, to explore further the molecular differences between (9+2) and (9+0) cilia, we examined the localisation of three highly conserved and core ODAD members – Odad1, Odad3 and Odad4 – to cilia in zebrafish embryos. Surprisingly, although all the Odad proteins exhibited axonemal localisation in (9+2) cilia, Odad1 and Odad3 localised exclusively to basal bodies of (9+0) cilia, whereas Odad4 was present along their axonemes.
Thus, we have clarified how two paralogous fMIPs, CFAP53 and MNS1, regulate (9+2) versus (9+0) cilia motility. Given that variants of the human orthologues have been implicated in ciliopathies with manifestation of male infertility and heterotaxy (Leslie et al., 2020; Narasimhan et al., 2015; Noel et al., 2016; Perles et al., 2012; Ta-Shma et al., 2018), our data provide a mechanistic framework for understanding the pathobiology of these disease phenotypes. Furthermore, localisation studies with ODAD members offer additional evidence for extensive molecular distinctiveness among motile cilia types. These findings inspire the need to evolve the cryo-EM technology so that, like (9+2) cilia, the architecture of other kinds of cilia can also be analysed at atomic resolution.
RESULTS
CFAP53 and MNS1 are paralogous fMIPs
To establish whether CFAP53 and MNS1 are paralogous, we studied their evolutionary relationships. Initially, we utilised MirrorTree software (Ochoa and Pazos, 2010) to determine their phylogenetic affinity. We observed that 59 organisms were common among the generated trees, and a correlation value of 0.895 was obtained with a P-value significance of ≤0.000001. Therefore, these proteins exhibit a strong co-evolutionary relationship. Although they share low sequence identity (22.8%) and similarity (48.9%), respectively (Fig. S1A), they are likely paralogues. We also prepared a phylogenetic tree to characterise their affinity. CFAP53 and MNS1 co-cluster, and are possibly paralogous in Ornithorhynchus anatinus, Sarcophilus harrisii, Monodelphis domestica, Rattus norvegicus and Mus musculus (Fig. S1B; Table S1).
We also performed remote homology searches using the sensitive profile-HMM homology detection method with human CFAP53 sequence as query. Indeed, this led to the identification of homologous regions in CFAP53 and MNS1. Specifically, we found an approximately 80-residue conserved region in CFAP53, adjoining the previously identified ELLEn module (Andersen et al., 2024) (Fig. 1E; Fig. S2A). This region, which we have termed the LRQ motif after its conserved residues, was also present in MNS1 and TRICHOPLEIN (TCHP), but not in another ELLEn module-containing MIP, CFAP141, suggesting that MNS1 is evolutionarily related to CFAP53 and TCHP. This LRQ motif is marked by an invariably conserved leucine-arginine (LR) signature motif across paralogues and species (Fig. 1E). Conversely, the ELLEn module is present only in CFAP53, TCHP and CFAP141, indicating that this region was lost in MNS1 (Fig. S2A).
Because CFAP53 and MNS1 structures have previously been determined in the PMDs of various organisms, including humans (Gui et al., 2021; Ma et al., 2019; Walton et al., 2023), we examined their spatial positions as 3D coordinates of PMDs from cilia of the respiratory epithelium [Protein Data Bank (PDB): 8J07]. Interestingly, this analysis revealed that the LR signature motif marks the topological site for the perpendicular kink of CFAP53 and MNS1, which disrupts the elongated structures of these proteins at their contact sites on the A tubule lattice, between protofilaments A6-A7 and A7-A8, respectively (Fig. S2B), further supporting their paralogous nature. Inspection of the 3D coordinates revealed that another fMIP, CFAP210, might also bear similarities to the LR signature motif (Fig. 1E; Fig. S2C,D). This observation was upheld by remote homology searches using CFAP53 as a query, which yielded matches to CFAP210 and CFAP45, albeit with low probability (E>0.01) (Fig. S2E). Because these proteins do not show significant homology to CFAP53 by sequence, we cautiously denote the inferred ‘LR’ motifs in CFAP210 and CFAP45 based on inspection of their 3D structures.
MNS1 localises to the ciliary axoneme of (9+0) and (9+2) cilia in zebrafish and mice
Immunolocalisation studies of endogenous or overexpressed epitope-tagged MNS1 in mouse sperm flagella, zebrafish LRO and human respiratory cilia as well as cryo-EM analysis have suggested that it is an axonemal protein (Choksi et al., 2014; Gui et al., 2021; Ma et al., 2019; Ta-Shma et al., 2018; Zhou et al., 2012). Nevertheless, to investigate whether, like CFAP53, MNS1 could localise to additional ciliary regions overlooked in these earlier studies, we used C-terminal haemagglutinin (HA)-tagged zebrafish Mns1 to visualise localisation to zebrafish KV and pronephric (kidney) duct cilia, which are (9+0) and (9+2) cilia-types, respectively. In both instances, we found that Mns1-HA protein localisation was restricted exclusively to ciliary axonemes (Fig. 2A,B). Given that Mns1-HA also rescued laterality defects of mns1 zebrafish mutants (see below), we argue that this pattern represents physiological localisation.
MNS1 localisation to motile cilia in zebrafish and mouse tissues. (A) Immunofluorescence with antibodies to HA epitope (green) and acetylated tubulin (magenta) of cilia in zebrafish KV [10-somite stage (14 hpf)]. (B) Immunofluorescence with antibodies to HA (green) and acetylated tubulin (magenta) of zebrafish pronephric duct (PD) cilia (24 hpf). (C) Immunofluorescence with antibodies to MNS1 (green) and acetylated tubulin (magenta) of E8.0 mouse node cilia. (D) Immunofluorescence with antibodies to MNS1 (green) and γ tubulin (magenta) of E8.0 mouse node cilia. (E) Immunofluorescence with antibodies to MNS1 (green) and acetylated tubulin (magenta) of adult mouse trachea. Insets in C-E show higher magnification views. Scale bars: 5 µm (A,B,E, main panel); 2 µm (C,D, main panels); 10 µm (C-E, insets).
MNS1 localisation to motile cilia in zebrafish and mouse tissues. (A) Immunofluorescence with antibodies to HA epitope (green) and acetylated tubulin (magenta) of cilia in zebrafish KV [10-somite stage (14 hpf)]. (B) Immunofluorescence with antibodies to HA (green) and acetylated tubulin (magenta) of zebrafish pronephric duct (PD) cilia (24 hpf). (C) Immunofluorescence with antibodies to MNS1 (green) and acetylated tubulin (magenta) of E8.0 mouse node cilia. (D) Immunofluorescence with antibodies to MNS1 (green) and γ tubulin (magenta) of E8.0 mouse node cilia. (E) Immunofluorescence with antibodies to MNS1 (green) and acetylated tubulin (magenta) of adult mouse trachea. Insets in C-E show higher magnification views. Scale bars: 5 µm (A,B,E, main panel); 2 µm (C,D, main panels); 10 µm (C-E, insets).
In mouse embryos, motile (9+0) cilia are found in the nodal pit at embryonic day (E) 8 (Nonaka et al., 1998). Using an antibody raised against human MNS1 (Ta-Shma et al., 2018), we could localise mouse MNS1 to nodal cilia axonemes (Fig. 2C; see Fig. S3 for antigen sequence conservation between mouse and human MNS1 proteins). Co-staining for MNS1 and γ-tubulin (Fig. 2D) indicated that MNS1 is also localised to the base of node cilia. However, this localisation pattern is somewhat different from that of CFAP53, which is found mainly at the node cilia base, the centriolar satellites (Ide et al., 2020). In mouse trachea, MNS1 localised to axonemes and the base of respiratory cilia (Fig. 2E), similar to CFAP53 (Ide et al., 2020).
Loss of Mns1 in zebrafish severely affects KV cilia motility
To investigate Mns1 function in ciliary motility, we first used the zebrafish to interrogate this issue. Our earlier work with the transcription factor Foxj1, a master regulator of motile cilia biogenesis, had identified mns1 as a target gene (Choksi et al., 2014). Consistent with this, mns1 is expressed in tissues that differentiate motile cilia during zebrafish embryogenesis (https://zfin.org/ZDB-GENE-030521-42/expression). We used CRISPR/Cas9 gene editing to introduce a 19 bp deletion within exon 3 of mns1 (mns1sq5722; Fig. 3A,B; see also Materials and Methods and Table S2). Conceptual translation of the mutant cDNA predicted a highly truncated protein, likely to be completely nonfunctional (Fig. 3C; also see Materials and Methods). Despite this, we failed to observe canonical cilia defect-associated anomalies among embryos from heterozygous fish crosses, such as axial curvature (which arises from ciliary defects within brain ventricles and spinal canal) or cystic kidneys (from cilia defects within kidney tubules). However, a significant proportion of the embryos (about 12.5%, i.e. half of the zygotic mutants) exhibited aberrant heart laterality, signifying LRO cilia defects (Fig. 3D). Given that mns1 mRNA is deposited maternally (Fig. S4A,B), we reasoned that stronger ciliary defect-associated phenotypes could be apparent in maternal-zygotic (mz) mutants. Therefore, we raised homozygous zygotic mutants to adulthood and obtained pure clutches of mz mutants by in-crossing. In these embryos, we noted the same heart laterality defects, but now in a much higher proportion (approximately half of every clutch) (Fig. 3D), and visualisation of KVs of these mz mutants confirmed strongly dysmotile cilia (Movies 1, 2). We could rescue the heart laterality defects using microinjection of sense mRNA encoding Mns1-HA (Fig. 3E), confirming mutation specificity and physiologically relevant localisation of Mns1-HA to ciliary axonemes. However, like the zygotic mutants, the mz mutants did not exhibit axial curvature or kidney cysts, and video microscopy of (9+2) cilia of multiciliated cells (MCCs) of the olfactory pits showed a wild-type beating pattern (Movies 3, 4). These findings clarify that, even though mns1 mRNA is deposited maternally, it does not contribute to masking of ciliary defects in the zygotic mutants, and the normal motility of (9+2) cilia and absence of phenotypic abnormalities, such as axial defects and kidney cysts, could be due to redundancy with Cfap53.
mns1 mutant zebrafish and L-R asymmetry defects. (A) Schematic of the zebrafish mns1 locus and targeting of exon 3 with guide RNA. (B) Nucleotide sequence of the mns1 mutant allele compared with wild type. (C) Schematic and amino acid sequence of the predicted Mns1 mutant protein compared with wild type. (D) Percentage of embryos (48 hpf) showing a heart looping defect compared with wild type (three technical replicates). (E) Percentage of mns1 mutant embryos (48 hpf) showing a heart looping defect after rescue with mns1-HA mRNA, compared with uninjected mns1 mutant embryos (three technical replicates). Data are presented as mean±s.d. *P≤0.05, **P≤0.01 (two-tailed Student's t-test).
mns1 mutant zebrafish and L-R asymmetry defects. (A) Schematic of the zebrafish mns1 locus and targeting of exon 3 with guide RNA. (B) Nucleotide sequence of the mns1 mutant allele compared with wild type. (C) Schematic and amino acid sequence of the predicted Mns1 mutant protein compared with wild type. (D) Percentage of embryos (48 hpf) showing a heart looping defect compared with wild type (three technical replicates). (E) Percentage of mns1 mutant embryos (48 hpf) showing a heart looping defect after rescue with mns1-HA mRNA, compared with uninjected mns1 mutant embryos (three technical replicates). Data are presented as mean±s.d. *P≤0.05, **P≤0.01 (two-tailed Student's t-test).
Mns1 knockout phenotype in mice
We generated Mns1 mutant mice, lacking exon 3 (Mns1em1Hmd, hereafter referred to as Mns1−/−), and predicted that this would result in a severely truncated protein (Fig. 4A-D; also see Materials and Methods). A previously published Mns1 null allele, which targeted exons 3-8, also resulted in a similar truncated protein (Zhou et al., 2012). About 30% of Mns1−/− mice died at birth, and another 40% succumbed within 12 weeks (Fig. 4E). Mutant mice that perished between 30 and 60 weeks exhibited hydrocephalus (3/3 mice). When 12 Mns1−/− mice were examined for laterality defects either at birth or later, 9/12 showed anomalies (Table 2). Two of them showed situs inversus totalis; the remaining seven showed a variable degree of heterotaxy. Abnormal lung lobation (9/12), reversed arching of the aorta (8/12), heart apex on the right (8/12) and stomach on the right (6/11) were frequently observed (Table 2). Moreover, Mns1−/− males were infertile (2/2), whereas Mns1−/− females were fertile (2/2).
ODA loss from Mns1 mutant mouse (9+0) nodal cilia. (A) Genetic structure of the wild-type Mns1 locus and generation of the knockout allele lacking exon 3. (B) Nucleotide sequence of the Mns1 mutant allele compared with wild type. (C) Schematic and amino acid sequence of the predicted MNS1 mutant protein compared with wild type. (D) PCR-based genotyping of Mns1−/− mice. (E) Survival curve of Mns1−/− mice. Mice analysed were 0-12 weeks of age, without sex bias. (F,G) TEM analysis of wild-type and Mns1–/– node cilia. Higher-magnification views of corresponding doublet microtubules (arrowheads), together with schematic diagrams of PMDs and ODAs (blue protrusions) are shown. Scale bars: 100 nm. (H) ODA number per node cilium from wild-type and Mns1−/− embryos. Data are presented as a violin plot. ****P≤0.0001 (two-tailed Student's t-test).
ODA loss from Mns1 mutant mouse (9+0) nodal cilia. (A) Genetic structure of the wild-type Mns1 locus and generation of the knockout allele lacking exon 3. (B) Nucleotide sequence of the Mns1 mutant allele compared with wild type. (C) Schematic and amino acid sequence of the predicted MNS1 mutant protein compared with wild type. (D) PCR-based genotyping of Mns1−/− mice. (E) Survival curve of Mns1−/− mice. Mice analysed were 0-12 weeks of age, without sex bias. (F,G) TEM analysis of wild-type and Mns1–/– node cilia. Higher-magnification views of corresponding doublet microtubules (arrowheads), together with schematic diagrams of PMDs and ODAs (blue protrusions) are shown. Scale bars: 100 nm. (H) ODA number per node cilium from wild-type and Mns1−/− embryos. Data are presented as a violin plot. ****P≤0.0001 (two-tailed Student's t-test).
As suggested by the laterality defects, node cilia of Mns1−/− embryos were almost completely immotile (Movies 5, 6). Transmission electron microscopy (TEM) analysis showed a complete absence of ODAs (Fig. 4F-H). Tracheal cilia of Mns1−/− mice were motile, but showed an abnormal beating pattern (Movies 7, 8). In particular, the beating angle was smaller, rendering the effective stroke inefficient, similar to what we have previously reported for Cfap53 mutants (Ide et al., 2020).
Concomitant loss of Mns1 and Cfap53 affects motility of zebrafish (9+2) cilia
Our previous work with antisense morpholinos and the study of Noel et al. with cfap53 mutant zebrafish had established that the protein is essential for KV cilia motility (Narasimhan et al., 2015; Noel et al., 2016). To examine potential redundant roles of Cfap53 and Mns1, we generated an independent cfap53 allele using CRISPR/Cas9 gene editing (Fig. 5A,B). This allele consists of a 14 bp deletion in exon 2 (cfap53sq5723), predicted to severely truncate the protein [Fig. 5B,C; the allele described by Noel et al. (2016) is a 7 bp deletion within the same exon; see also Materials and Methods]. cfap53 mRNA is not deposited maternally (Fig. S4C,D), and we raised embryos obtained from heterozygous in-crosses to adulthood. In-cross of homozygous mutant fishes yielded mz mutants, which did not exhibit any other ciliary abnormalities besides KV cilia motility defects (Movie 9) and abnormal L-R asymmetry (Fig. 5D), recapitulating the severe loss of KV cilia motility and laterality defects reported by Noel et al. as well as in our study with cfap53 morpholino (Narasimhan et al., 2015; Noel et al., 2016). We also examined the motility of (9+2) MCC cilia of olfactory pits and could not discern obvious defects (Movie 10). Thus, like Mns1, Cfap53 is dispensable for zebrafish (9+2) cilia motility, again pointing to a likely redundancy with Mns1.
Severe ciliary disorder in zebrafish cfap53; mns1 double mutants. (A) Schematic of the zebrafish cfap53 locus and targeting of exon 2 with guide RNA. (B) Nucleotide sequence of the cfap53 mutant allele compared with wild type. (C) Schematic and amino acid sequence of the predicted Cfap53 mutant protein compared with wild type. (D) Percentage of embryos showing a heart looping defect (48 hpf) compared with wild type (three technical replicates). Data are presented as mean±s.d. *P≤0.05 (two-tailed Student's t-test). (E) Wild-type (WT) zebrafish embryo morphology (24 hpf). (F) A cfap53−/−; mns1−/− double mutant embryo with curved body axis (24 hpf). Scale bars: 1 mm.
Severe ciliary disorder in zebrafish cfap53; mns1 double mutants. (A) Schematic of the zebrafish cfap53 locus and targeting of exon 2 with guide RNA. (B) Nucleotide sequence of the cfap53 mutant allele compared with wild type. (C) Schematic and amino acid sequence of the predicted Cfap53 mutant protein compared with wild type. (D) Percentage of embryos showing a heart looping defect (48 hpf) compared with wild type (three technical replicates). Data are presented as mean±s.d. *P≤0.05 (two-tailed Student's t-test). (E) Wild-type (WT) zebrafish embryo morphology (24 hpf). (F) A cfap53−/−; mns1−/− double mutant embryo with curved body axis (24 hpf). Scale bars: 1 mm.
To investigate this, we generated double-heterozygous mns1+/−; cfap53+/− fish, and in-crossed them to obtain zygotic double mutants. As expected from Mendelian genetics, we found that approximately 6.25% of embryos from such crosses exhibited prominent axial curvature (Fig. 5E,F), typical of zebrafish with strong cilia motility defects, that was not apparent in single mutants for either gene. Genotyping these embryos confirmed them to be double mutants. The double mutants also exhibited heart laterality defects, hallmark phenotype of the single mutants, with about 50% of the double mutants having the heart tube looped on the wrong side (Fig. 5D). Despite the axial curvature, the double mutants did not develop cystic kidneys, suggesting that (9+2) cilia motility may not be fully compromised. Consistent with this notion, when we analysed (9+2) cilia of olfactory pit MCCs, a range of motility defects was apparent. These cilia exhibited aberrant motion distinct from the smooth waveform characteristic of wild-type embryos (Movie 11). In a proportion of double mutants (about 20%), we found complete paralysis of the majority of these cilia, and those that were still motile showed uncoordinated twitching (Movie 12). Because mns1; cfap53 zygotic homozygotes are embryonic lethal owing to their curved body axis precluding swimming, we generated mns1−/−; cfap53+/− adults and bred them to obtain mz mutant mns1 and zygotic mutant cfap53 double homozygotes. These mutants exhibited a phenotypic spectrum and (9+2) olfactory pit cilia abnormalities identical to the zygotic mns1; cfap53 double mutants (data not shown). Because KV cilia are immotile or exhibit strong motility defects in cfap53 and mns1 single mutants, respectively, we did not analyse their motility in the double mutants.
(9+2) cilia of Mns1; Cfap53 double mutant mice are devoid of ODAs
To test whether Mns1 and Cfap53 function in distinct or similar pathways in mammalian cilia, we generated double mutant mice. First, mice lacking both MNS1 and CFAP53 function showed a much higher incidence of lethality after birth. Unlike Mns1 and Cfap53 single mutants, about 60% of the double mutants died at birth, and most of them perished within 5 weeks (Fig. 6A). Second, the double mutants exhibited more severe laterality defects than the single mutant mice (Fig. 6B,C; Table 3). In particular, the incidence of azygous vein and stomach on the right and abnormal liver lobation was higher in the double mutants. Finally, the tracheal cilia of Mns1; Cfap53 double mutants lost ODAs almost completely, whereas this effect was only partial in the single mutants (Fig. 6D-G; see also Ide et al., 2020). Owing to early lethality of the double mutants, we were unable to perform live imaging of tracheal ciliary beating as the few tracheae that we could isolate were used for TEM analysis.
ODA loss from Cfap53; Mns1 double mutant mouse (9+2) tracheal cilia. (A) Survival curve of Cfap53−/−; Mns1−/− double mutants. Mice analysed were 0-5 weeks of age, without sex bias. (B,C) Laterality defects of Cfap53−/−; Mns1−/− mice: heart apex (asterisk) in reversed (dextral) position (B), and spleen (double asterisk) in reversed (dextral) position (C). (D-F) Tracheal cilia TEM analysis from Cfap53+/+ (D), Mns1–/– (E) and Cfap53–/–; Mns1−/− mice (F). Insets show higher-magnification views of PMDs (arrowheads), together with their schematics including the ODAs (blue protrusions). Scale bars: 100 nm. (G) ODA number per tracheal cilium from wild-type control, Mns1−/− and Cfap53−/−; Mns1−/− mice. Data are presented as mean±s.d. ****P≤0.0001 (two-tailed Student's t-test).
ODA loss from Cfap53; Mns1 double mutant mouse (9+2) tracheal cilia. (A) Survival curve of Cfap53−/−; Mns1−/− double mutants. Mice analysed were 0-5 weeks of age, without sex bias. (B,C) Laterality defects of Cfap53−/−; Mns1−/− mice: heart apex (asterisk) in reversed (dextral) position (B), and spleen (double asterisk) in reversed (dextral) position (C). (D-F) Tracheal cilia TEM analysis from Cfap53+/+ (D), Mns1–/– (E) and Cfap53–/–; Mns1−/− mice (F). Insets show higher-magnification views of PMDs (arrowheads), together with their schematics including the ODAs (blue protrusions). Scale bars: 100 nm. (G) ODA number per tracheal cilium from wild-type control, Mns1−/− and Cfap53−/−; Mns1−/− mice. Data are presented as mean±s.d. ****P≤0.0001 (two-tailed Student's t-test).
CFAP53 interacts with MNS1, but unlike MNS1 does not self-associate
In our earlier work with CFAP53, we found that the protein can interact with the ODA docking complex member ODAD4 as well as several ODA dyneins (Ide et al., 2020). In addition, based on the localisation of CFAP53 to centriolar satellites and the established role of the satellites in facilitating protein transport into the axoneme (Prosser and Pelletier, 2020), we proposed that CFAP53 could also be involved in the transport of ODAD4 and ODA dyneins into cilia (Ide et al., 2020). In line with these interactions, ODAs as well as ODAD4 were lost from nodal cilia of Cfap53 mutant mice, but in (9+2) cilia there was only a partial ODA loss and ODAD4 localisation remained unaffected (Ide et al., 2020). Biochemical studies with MNS1 have also shown that it can interact with another ODA docking complex member, ODAD1 (Ta-Shma et al., 2018). Additionally, MNS1 has been shown to interact with itself, forming dimers and oligomers (Zhou et al., 2012). Indeed, end-to-end self-association for MNS1 as well as CFAP53 has been noted in cryo-EM studies, allowing them to generate a 48 nm periodicity along the axoneme (Gui et al., 2021; Ma et al., 2019). Moreover, the atomic models predict interaction of MNS1 with CFAP53 as well (Gui et al., 2021; Ma et al., 2019). To validate CFAP53 self-association and interaction of CFAP53 with MNS1, we performed immunoprecipitation studies with overexpressed epitope-tagged mouse proteins in cultured HEK293T cells. We observed that CFAP53-Myc can immunoprecipitate with MNS1-HA (Fig. 7A). However, unlike MNS1, we were unable to detect any self-association between CFAP53-Myc and CFAP53-HA (Fig. 7B). Consistent with this, although MNS1-HA and MNS1-Myc appeared to form oligomeric filaments in the cytoplasm, as reported previously (Zhou et al., 2012), we failed to observe such filaments with CFAP53 (Fig. 7C,D; also see Discussion).
Biochemical interaction between MNS1 and CFAP53. (A) Myc-tagged MNS1 immunoprecipitated with HA-tagged CFAP53. (B) MNS1, but not CFAP53, exhibited self-interaction. (C) MNS1-Myc and MNS1-HA formed filament-like structures in HEK293T cells (arrows). MNS1-HA was stained with anti-HA antibody (magenta); MNS1-Myc was stained with anti-Myc antibody (cyan). (D) CFAP53-Myc and CFAP53-HA showed diffuse staining in HEK293T cells. CFAP53-HA was stained with anti-HA antibody (magenta); CFAP53-Myc was stained with anti-Myc antibody (Cyan). Scale bars: 5 µm. Immunoprecipitation was verified with two biological replicates. Immunofluoresence data represent two technical replicates with 100 cells analysed per replicate.
Biochemical interaction between MNS1 and CFAP53. (A) Myc-tagged MNS1 immunoprecipitated with HA-tagged CFAP53. (B) MNS1, but not CFAP53, exhibited self-interaction. (C) MNS1-Myc and MNS1-HA formed filament-like structures in HEK293T cells (arrows). MNS1-HA was stained with anti-HA antibody (magenta); MNS1-Myc was stained with anti-Myc antibody (cyan). (D) CFAP53-Myc and CFAP53-HA showed diffuse staining in HEK293T cells. CFAP53-HA was stained with anti-HA antibody (magenta); CFAP53-Myc was stained with anti-Myc antibody (Cyan). Scale bars: 5 µm. Immunoprecipitation was verified with two biological replicates. Immunofluoresence data represent two technical replicates with 100 cells analysed per replicate.
Distinct localisation of ODAD members to motile cilia-types in zebrafish
Besides CFAP53 and MNS1, which have distinct localisation and/or function in (9+0) versus (9+2) cilia, we have previously shown that, in mice, individual ODA dyneins also have unique expression patterns, localisation and function in these two cilia types. Whereas DNAH11 is localised along the entire length of LRO cilia axonemes, it is localised specifically to the proximal region of tracheal cilia (Ide et al., 2020). Even more intriguingly, the DNAH9 protein, which localises to the distal axoneme of tracheal cilia, is not expressed in the node despite the gene being transcribed in node cells (Ide et al., 2020). These observations illustrate that, besides the traditional view that the absence of the central pair and associated structures like radial spokes is the key defining difference between (9+0) and (9+2) cilia, several other ciliary components have differential expression, localisation and/or function in the two cilia types. Because cryo-EM level ultrastructural studies with (9+0) cilia are currently not possible, we decided to explore differences between (9+0) and (9+2) cilia by evaluating localisation patterns of a set of evolutionarily conserved and key motility components – ODAD members Odad1, Odad3 and Odad4 – to zebrafish motile cilia. We used an N-terminal Myc tag for all three proteins, and for the C terminus, we used GFP for Odad1 and Odad3, and an HA tag for Odad4. We injected synthetic sense mRNAs for these tagged proteins into zebrafish eggs and analysed their localisation to cilia in KV, kidney ducts and spinal canal; TEM analysis of the latter has revealed them to be largely of the (9+0) type (Kramer-Zucker et al., 2005; Sedykh et al., 2016). For Odad1 and Odad3, N-terminal Myc-tagged versions localised almost exclusively to axonemes of (9+2) kidney duct cilia, consistent with reports of localisation of the endogenous proteins with specific antibodies to (9+2) airway cilia of mammals (Hjeij et al., 2014; Onoufriadis et al., 2013; Wallmeier et al., 2016) (Fig. 8A,B). However, we failed to observe any ciliary localisation of Myc-Odad4 (data not shown). Similarly, C-terminal GFP-tagged Odad1 and Odad3 as well as C-terminal HA-tagged Odad4, localised along kidney cilia axonemes, consistent with previous overexpression studies with C-terminally tagged versions of these proteins to zebrafish and Xenopus (9+2) cilia (Hayes et al., 2007; Jerber et al., 2014) (there was also a substantial localisation of Odad4-HA in the cytoplasm) (Fig. 8C; Fig. S5A,B). By contrast, N- as well as C-terminally tagged Odad1 and Odad3 localised exclusively to the base of KV well as spinal canal cilia, whereas for Odad4, localisation of HA-tagged C-terminal version was observed along their axonemes, similar to our previous report for (9+0) mouse LRO cilia (Ide et al., 2020) (Fig. 8D-I; Fig. S5C-F). There was also substantial Odad4-HA accumulation in the cytoplasm, and we again failed to observe localisation of Myc-Odad4 to these cilia (data not shown; see Discussion on why an N-terminal tag could interfere with Odad4 localisation).
Differential localisation of Odad proteins to zebrafish motile cilia axonemes and basal bodies. (A-C) Odad1, Odad3 and Odad4 localised along axonemes of (9+2) cilia in zebrafish pronephric duct (PD) at 24 hpf. (D,E) Odad1 and Odad3 localised to KV cilia base at the 10-somite stage. (F) Odad4 localised along axonemes of KV cilia. (G,H) Odad1 and Odad3 localised at the spinal canal (SC) cilia base at 24 hpf. (I) Odad4 localised along axonemes of SC cilia at 24 hpf. Odad proteins were detected with anti-Myc antibody (for Odad1 and Odad3; cyan, arrows) or anti-HA antibody (for Odad4; cyan, arrows), cilia with anti-acetylated tubulin antibody (magenta) and nuclei with DAPI (blue). (J) An odad3 morphant (MO) (48 hpf), showing a ventrally curved body axis. (K) An odad3 morphant (MO) (48 hpf), with rescue of axial curvature on co-injection of odad3-gfp mRNA. Data represent three technical replicates with n=75 per replicate. Scale bars: 5 µm (A-I); 1 mm (J,K).
Differential localisation of Odad proteins to zebrafish motile cilia axonemes and basal bodies. (A-C) Odad1, Odad3 and Odad4 localised along axonemes of (9+2) cilia in zebrafish pronephric duct (PD) at 24 hpf. (D,E) Odad1 and Odad3 localised to KV cilia base at the 10-somite stage. (F) Odad4 localised along axonemes of KV cilia. (G,H) Odad1 and Odad3 localised at the spinal canal (SC) cilia base at 24 hpf. (I) Odad4 localised along axonemes of SC cilia at 24 hpf. Odad proteins were detected with anti-Myc antibody (for Odad1 and Odad3; cyan, arrows) or anti-HA antibody (for Odad4; cyan, arrows), cilia with anti-acetylated tubulin antibody (magenta) and nuclei with DAPI (blue). (J) An odad3 morphant (MO) (48 hpf), showing a ventrally curved body axis. (K) An odad3 morphant (MO) (48 hpf), with rescue of axial curvature on co-injection of odad3-gfp mRNA. Data represent three technical replicates with n=75 per replicate. Scale bars: 5 µm (A-I); 1 mm (J,K).
To adduce evidence that these localisation patterns of tagged and overexpressed Odad proteins indeed reflect their endogenous localisation patterns and do not compromise function, we attempted to rescue odad3 morphants with the N-terminal Myc and C-terminal GFP chimeric proteins. As described previously, odad3 morphants exhibit ventrally curved body axis and defects in heart looping (Jerber et al., 2014), closely mimicking odad3 genetic mutants (Hjeij et al., 2014), and we were able to reproduce these phenotypes after injecting a splice-blocking morpholino used by Jerber et al. (2014) into wild-type zebrafish eggs (Fig. 8J). Analysis of KV cilia motility of the morphants revealed immotile or highly dysmotile cilia, in line with the highly penetrant heart laterality defects (Movie 13). Notably, mRNAs for both N- as well as C-terminally tagged Odad3 were able to significantly rescue KV cilia motility, axial curvature and heart situs defects of the morphants (Fig. 8K; Movie 14), confirming that the ciliary localisation patterns exhibited by the chimeric Odad3 proteins are physiological, and the epitope tags do not disrupt their biological activity.
DISCUSSION
We and others first identified CFAP53 as a ciliary protein based on screens aimed at discovering novel ciliary/centriolar components (Choksi et al., 2014; Firat-Karalar et al., 2014; Hayes et al., 2007; Hoh et al., 2012; Ross et al., 2007). Subsequently, using zebrafish knockdown studies and genetic analysis of a human individual with situs inversus, we established that CFAP53 dysfunction is associated with ODA loss from LRO cilia, rendering them immotile (Narasimhan et al., 2015). These findings were independently validated by other groups, thereby confirming CFAP53 as a key protein responsible for proper LRO cilia motility and establishment of L-R asymmetry (Noel et al., 2016; Perles et al., 2012; Silva et al., 2016). Moreover, investigating CFAP53 localisation led us and others to conclude that it is an axonemal protein and is also associated with ciliary basal bodies, likely as a centriolar satellites constituent (Narasimhan et al., 2015; Silva et al., 2016). More recently, we investigated CFAP53 localisation and function in mice (Ide et al., 2020). In line with zebrafish data, we found that CFAP53 is a motile cilia protein with differential localisation in (9+0) versus (9+2) cilia, present mainly in basal bodies of the former and in basal bodies as well as along axonemes of the latter. Moreover, CFAP53 loss preferentially affected (9+0) cilia motility, with (9+2) cilia altered marginally in their beat frequency and amplitude. Function of MNS1, another fMIP, was originally investigated based on its prominent expression in the mouse testis, specifically in spermatocytes, spermatids and along axonemes of mature sperm (Zhou et al., 2012). Mutant mice developed rudimentary sperm flagella, with highly disorganised ultrastructure. In addition, these mice exhibited randomised L-R asymmetry and hydrocephalus, with partial ODA loss from tracheal (9+2) cilia (Zhou et al., 2012). However, ciliary motility defects were not investigated. Like CFAP53, mutation of MNS1 is also linked with situs abnormalities and male infertility in humans (Leslie et al., 2020; Ta-Shma et al., 2018). Also, like CFAP53, which can interact with ODA dyneins as well as the ODAD member ODAD4, MNS1 associates with the ODAD protein ODAD1 (Ta-Shma et al., 2018).
However, it is cryo-EM analyses of (9+2) cilia which have revealed that CFAP53 and MNS1 are fMIPs, decorating lumens of the A microtubule protofilament (Gui et al., 2021; Ma et al., 2019). CFAP53 and MNS1 occupy clefts between adjacent protofilaments – CFAP53 between protofilaments A6 and A7 and MNS1 between A7 and A8. Because of this neighbouring localisation, interactions with ODAD components and ODA loss when the proteins are mutated led us to investigate whether they have redundant roles in regulating (9+2) cilia motility. We established that CFAP53 and MNS1 are paralogues, and we also identified a conserved region, the LRQ motif, in CFAP53, MNS1 (and another ciliary protein, TCHP). This motif features a conserved LR signature and marks a critical site in CFAP53 and MNS1 structure, affecting their configuration within the A tubule. Consistent with this, in mice and zebrafish doubly mutant for both proteins, (9+2) cilia motility is strongly compromised with almost complete ODA loss.
In addition, we have performed a detailed evaluation of cilia structure and motility in the Mns1 single mutants, which were not examined previously (Zhou et al., 2012). Although LRO cilia motility in zebrafish as well as mice was strongly affected, (9+2) cilia remained motile, with a partial loss of ODAs (in mice). These data collectively establish that CFAP53 and MNS1 are both essential for LRO cilia motility, but their individual functions are largely dispensable in (9+2) cilia. We propose that this disparity likely arises from their distinct localisation patterns and activities in the two kinds of cilia. In (9+0) LRO cilia, CFAP53 possibly facilitates transport of ODA dyneins and ODAD members as part of the centriolar satellites, whereas MNS1 functions as an fMIP, stably anchoring ODADs and ODAs to the axoneme. By contrast, in (9+2) cilia, both proteins predominantly function as fMIPs for ODAD and ODA attachment, resulting in a considerable degree of redundancy. An intriguing observation that we made with zebrafish mns1 and cfap53 mutants is that, even though cilia within the spinal canal, which drive cerebrospinal fluid flow and formation of the glycoproteinaceous Reissner fibre for development of a straight body axis (Roy, 2021), have been described to be (9+0) (Kramer-Zucker et al., 2005; Sedykh et al., 2016), no axial defects were apparent in the single mutants. This suggests that (9+0) KV cilia are distinct from those in the spinal canal with respect to their requirement for Mns1 or Cfap53. Likewise, even though we have organised our narrative around LRO cilia being (9+0), in reality they are a mixture of (9+0) and (9+2), in zebrafish as well as mice (Caspary et al., 2007; Gui et al., 2021; Pinto et al., 2021) (the exact proportion of the two cilia types is presently not clear). However, most LRO cilia exhibit rotational motion (apparent in Movies 1, 5), ciliary components, such as CFAP53, localise exclusively to LRO cilia base, and mutation of Cfap53 and Mns1 uniformly affects their motility, implying that LRO (9+2) cilia could be distinct from those in other tissues. These findings categorically establish that motile cilia in different tissues of the vertebrate body are considerably more diverse than presently recognised.
Cryo-EM data have also revealed that both CFAP53 and MNS1 associate in a head-to-tail fashion, forming a 48 nm repeat long the A tubule (Gui et al., 2021; Ma et al., 2019). For MNS1, overexpression as well as biochemical studies, have confirmed oligomerisation (Zhou et al., 2012 and our present study). By contrast, we were unable to demonstrate self-association of CFAP53 by immunoprecipitation of overexpressed proteins. However, co-immunoprecipitation conditions may not be congenial for preserving native interactions (especially relatively weak interactions), and CFAP53 self-association is plausible in the axonemal microtubule environment. In addition to self-association, cryo-EM data also indicate interaction between the two proteins, which we have substantiated. Furthermore, each protein has been proposed to associate with a complement of various other MIPs, unique to each protein. These include CFAP21 (EFHB), C1ORF158 (CFAP107) and CFAP161 for MNS1 and PIERCE1, PIERCE2, and NME7 for CFAP53 (Gui et al., 2021; Ma et al., 2019). Validity and significance of these associations will require further biochemical studies.
Similar to CFAP53 and MNS1, loss of function of several other MIPs, including CFAP45, CFAP52, ENKUR and NME7, in humans manifest in strong L-R asymmetry defects (signifying LRO cilia dysmotility), but have a relatively mild effect on respiratory (9+2) cilia (Dougherty et al., 2020; Reish et al., 2016; Sigg et al., 2017; Ta-Shma et al., 2015). Although the reason for this disparity is presently unclear, these data nevertheless provide additional support that (9+0) and (9+2) cilia have key structural differences. In the final section, we attempted to further probe the molecular differences between (9+0) and (9+2) cilia. Rather surprisingly, we found that even highly evolutionarily conserved core ODAD members localised differentially to KV, spinal canal and kidney tubule cilia in zebrafish. Like Cfap53, Odad1 and Odad3 localised exclusively to KV and spinal canal cilia base (consistent with previous observations of Odad3 localisation to spinal canal cilia base; Jerber et al., 2014), raising the intriguing possibility that, in addition to their role in anchoring dynein arms to axonemes, they could have other functions at the ciliary base. This could be dynein arm transport, as we have invoked for CFAP53, or roles that are presently not defined. Exploring the interactome of these Odad proteins could provide clues for these additional functions. Also, absence of Odad1 and Odad3, central components of the pentameric vertebrate ODAD, from KV and spinal canal cilia axonemes suggests a significantly simplified ODAD or substitution by other proteins. For Odad4, we failed to observe ciliary localisation with an N-terminal Myc tag. The first ten N-terminal amino acids ODAD4 are not resolved in cryo-EM structure, indicating that this region is flexible, and an N-terminal tag should not interfere with ODAD4 binding to microtubules. One possibility is that the N-terminal Myc tag affected Odad4 transport into cilia. Alternatively, the anti-Myc antibody could have limited access to the N-terminal tag because the N terminus of ODAD4 is located in a small space between axonemal microtubules and ODA motor domain (M. Gui, personal communication). Unfortunately, whether ODAD1 and ODAD3 also localise to mouse LRO cilia base could not be determined owing to the lack of specific antibodies or transgenes expressing epitope-tagged proteins.
In conclusion, our work establishes that, although ultrastructural data of (9+2) cilia have been key to our understanding of their molecular composition and motility, such information cannot be directly transposed to (9+0) cilia, and to a variety of other motile cilia subtypes. Until we can obtain high-resolution structural data for different kinds of cilia, genetics and cell biological investigations, as illustrated here, will continue to remain instrumental in enabling us to dissect their molecular composition and mechanisms of motility. How such cilia diversity is assembled in different tissues (in some instances even within the same tissue) and how molecular differences engender unique motility profiles that drive fluid flow patterns suited for specific developmental and physiological requirements, promises to be an interesting area for future investigations.
MATERIALS AND METHODS
Protein orthologue or paralogue identification and phylogenetic analysis
Protein sequences of human MNS1 and CFAP53 were obtained from the UniProt database (UniProt, 2019). MNS1 and CFAP53 were aligned using EMBOSS needle (Needleman and Wunsch, 1970; Rice et al., 2000) for the computation of sequence identities between them. Orthologues or similar sequences were determined using domain enhanced lookup time accelerated BLAST (DELTA-BLAST) (Boratyn et al., 2012). Sequences sharing an E-value of ≤1e−04, query length coverage of ≥70% and sequence identity of ≥45% with the query proteins from the NCBI non-redundant database (Pruitt et al., 2007) were taken as close orthologues. MNS1 and its orthologues (244 sequences) and CFAP53 and its orthologues (87 sequences) were aligned with an iterative refinement method (MAFFT-L-INS-I version 7) giving preference to local alignment such that the conserved regions were well aligned (Katoh et al., 2002; Katoh and Standley, 2013). Further, multiple sequence alignment (MSA) of orthologous MNS1 and CFAP53 proteins were analysed with the help of the MirrorTree program (Ochoa and Pazos, 2010). Organisms in which MNS1 and CFAP53 shared an evolutionary relationship were selected, and a combined MSA and phylogenetic tree were prepared to determine whether these genes could be paralogous. The phylogenetic tree was generated in the Randomized Axelerated Maximum Likelihood (RAxML version 8.1.6) (Stamatakis, 2014) program, which utilises the maximum likelihood algorithm. Phylogenetic trees were visualised in FigTree (http://tree.bio.ed.ac.uk/software/figtree/). Additionally, selected MNS1 and CFAP53 protein and gene sequences were aligned with the help of EMBOSS Matcher (Rice et al., 2000; Stamatakis, 2014).
Remote homology searches
To assess potential homologies between human CFAP53 and other human proteins, we performed remote homology searches using the profile-HMM method with human CFAP53 sequence as a query in an initial HHblits search (Remmert et al., 2011) against the UniRef30 database, with eight iterative searches as default. An E-value cutoff of E=0.01 was applied for MSA generation. The resulting sequences were utilised to construct MSAs through multiple reiterative searches. Constructed MSAs were used to build hidden Markov models (HMMs) suitable for iterative profile-to-profile searches (profile-HMM) with the open-source software package HH-suite (https://github.com/soedinglab/hh-suite) and HHpred to retrieve remote homologous sequences in the protein PFAM database of hidden Markov models. Only local realignment search modes were used. These initial searches identified homologies in the CFAP53 N terminus to human TCHP N terminus and CFAP141, as previously shown (Andersen et al., 2024). Reassessing CFAP53 matches to TCHP and CFAP141 using the extracted CFAP53 N terminus (residue 9-119) as query in a second round of profile-HMM searches produced matches to only CFAP141 and TCHP (besides CFAP53) with E-values 2E−10 and 8E−06, respectively. Interestingly, using a slightly longer CFAP53 N-terminal search query (residue 9-190) additionally identified human MNS1 as a match (E-value 0.0016), suggesting a hitherto unknown homology between CFAP53 and MNS1, downstream of the previously identified ELLEn module. Indeed, using a short CFAP53 region (residue 108-188) downstream of the CFAP53 ELLEn module in a final profile-HMM search, matched with both human TCHP and MNS1 (E-values 1.1E−09 and 1.1E−05, respectively), but not with CFAP141, indicating that this conserved region in CFAP53 is shared with TCHP and MNS1. To confirm the remote homologies identified in the initial (forward) searches, the matching MNS1 sequence was extracted and used as a query for reciprocal profile-HMM searches. These reciprocal searches reaffirmed the homologies with CFAP53 and TCHP, establishing that MNS1 is a paralogue of the fMIP family, which includes CFAP53 and CFAP141, as well as TCHP.
3D structural assessment
Coordinates of entire cilia PMDs from all studies providing PDB structure results (PDB: 7UNG, 8J07) were retrieved from the PDB (https://www.rcsb.org/). Structures were analysed in BIOVA Discovery Studio Visualizer (BIOVIA, Dassault Systèmes, v4.5, San Diego: Dassault Systèmes, 2023) or PyMol (https://pymol.org/).
Zebrafish housing and strains
All zebrafish strains were maintained according to established protocols for fish husbandry at the Institute of Molecular and Cell Biology zebrafish facility. The following wild-type and mutant strains were used in this study: AB (inbred wild-type control), mns1sq5722 mutant (19 bp deletion in exon 3) and cfap53sq5723 mutant (14 bp deletion in exon 2) (generated using CRISPR/Cas9 technology; for guide RNA sequences and genotyping primers, see Table S2). All experiments with the zebrafish were approved by the Singapore National Advisory Committee on Laboratory Animal Research (protocol number: 221702).
Generation of cfap53 and mns1 mutant zebrafish strains
Single guide RNAs (sgRNAs) for the cfap53 or mns1 gene were designed using the web tool CHOPCHOP (https://chopchop.cbu.uib.no/). The target sites of sgRNAs were selected by identifying sequences that correspond to GGN(18)NGG in the DNA. sgRNAs were synthesised based on a modified two-component system (Bassett et al., 2013). In brief, PCR was conducted to generate a DNA template containing the T7 polymerase binding site, the sgRNA target sequence, along with a common reverse oligonucleotide of sgRNA sequence. Phusion polymerase (NEB, M0530S) was employed for this process. Subsequently, in vitro transcription of sgRNA was carried out using 500 ng of purified DNA template, with the MEGAshortscript T7 kit (Ambion, AM1354 M). One nanolitre of a mixture of 700 ng of the Cas9 protein (Toolgen) and 700 ng of each sgRNA was injected into one-cell-stage zebrafish embryos to generate F0 mutant fish. F0 fish were outcrossed to wild type, and transmitting founders were identified from embryo PCR and sequencing. Heterozygous F1 fish with desired mutations were used to establish the mutant strains. Three isoforms of the mns1 and three isoforms of cfpa53 transcripts are reported in the Ensembl database, and the induced mutations are predicted to cause severe truncation of proteins encoded by all transcript variants. Double mutant embryos were generated using standard genetic crosses of double heterozygous mutants.
Generation of odad3 morphants and rescue with odad3 mRNA
Wild-type zebrafish embryos were injected at the one-cell stage with 0.5 nl of a 1 µM solution of odad3 splice blocking morpholino (Table S2) (Jerber et al., 2014). For rescue of odad3 morphants, 0.5 nl of a 250 µg/µl solution of myc-odad3 or odad3-gfp mRNA was injected into wild-type zebrafish eggs following injection of the odad3 morpholino. Analysis of KV cilia motility, body axis curvature and heart looping was conducted at the 10-somite, 24 hours post-fertilisation (hpf) and 48 hpf stages of development of the odad3 morphants, respectively.
Mouse housing and strains
Mice were maintained at the animal facility of RIKEN Centre for Biosystems Dynamics Research (BDR), Japan, under a 12 h light-12 h dark cycle, and were provided with food and water ad libitum. All experiments were conducted in the C57BL/6 background, approved by the Institutional Animal Care and Use Committee (permission number: A2016-01-6), and carried out in accordance with guidelines of the RIKEN BDR. Mns1 knockout mice, lacking exon 3, were generated with the CRISPR/Cas9 system. Three transcript isoforms of Mns1 are reported in the Ensembl database, and the mutation in exon 3 is predicted to cause severe truncation of the proteins encoded by all transcript variants. Construction of the Cfap53 mutant mouse strain has been described previously (Ide et al., 2020).
Antibodies for immunofluorescence
Primary antibodies used in the mouse study were: mouse anti-acetylated tubulin (Sigma-Aldrich, T6793, 1:2000), mouse anti-γ-tubulin (Sigma-Aldrich, T5326, 1:1000) and rabbit anti-MNS1 (Sigma-Aldrich Prestige Antibodies, HPA039975, 1:500). Antibodies used in the zebrafish study were: rabbit anti-HA (Santa Cruz Biotechnology, SC805, 1:500), rabbit anti-Myc (Santa Cruz Biotechnology, SC289, 1:500), mouse anti-HA (Santa Cruz Biotechnology, SC7392, 1:500), rabbit anti-GFP (Torrey Pines Biolabs, TP401, 1:1000), mouse anti-acetylated-tubulin (Sigma-Aldrich, T6793, 1:500) and mouse anti-γ-tubulin (Sigma-Aldrich, T6557, 1:500). Alexa Fluor-conjugated secondary antibodies for immunofluorescence staining were purchased from Invitrogen [Alexa 488 goat anti-chicken (A-11039), Alexa 488 goat-anti mouse (A-11029), Alexa 488 goat anti-rabbit (A-11034), Alexa 555 goat anti-rabbit (A-21428) and Alexa 555 goat anti-mouse (A-28180); used at 1:500 for zebrafish experiments and 1:1000 for mouse experiments].
Immunofluorescence staining of mouse embryos and tracheae
Immunofluorescence staining of node cilia of E8.0 embryos and tracheal cilia was performed as described previously (Ide et al., 2020). All antibody-labelled samples were analysed with Olympus FV1000 and FV3000 confocal microscope systems.
Immunofluorescence staining of zebrafish embryos
Embryos were fixed for 2 h at room temperature using fish fix [4% (w/v) paraformaldehyde and 4% (w/v) sucrose in PBS base]. Subsequently, the fixed embryos were stored in methanol at −20°C. For immunofluorescence studies, embryos underwent a series of washes in a decreasing methanol:PBS gradient, followed by PBS washes and blocking in PBDT [1% (w/v) bovine serum albumin, 1% DMSO, 0.5% Triton X-100, PBS base] for 1 h. Primary antibodies were introduced into PBDT and incubated with the embryos overnight at 4°C. Subsequently, embryos were washed in PBDT before incubating with fluorophore-conjugated secondary antibodies and DAPI for 3 h at room temperature. Stained embryos were cleared in 70% glycerol, mounted, and imaged using an Olympus FV3000 confocal microscope. Image acquisition and analysis were performed using Olympus Fluoview FV10-ASW software.
Immunofluorescence analysis of HEK293T cells with MNS1 and CFAP53 overexpression
HEK293T cells (obtained from ATCC and tested free of contamination) were split and grown on glass coverslips in six-well plates one day before transfection. For transfection, 2 µg MNS1-HA/2 µg MNS1-Myc and 2 µg CFAP53-HA/2 µg CFAP53-Myc plasmids (encoding the mouse CFAP53 and MNS1 proteins) were incubated with 20 µl of Lipofectamine 2000 (Invitrogen) in 300 µl Optimal MEM (Invitrogen) for 5 min and added into the cell culture medium; 18 h post-transfection, the transfected cells were fixed with 4% paraformaldehyde in PBS base for 30 min. Primary antibodies (rabbit anti-HA and mouse anti-Myc) were used to stain the tagged proteins (both used at 1:200). The stained cells were mounted and imaged using an Olympus FV3000 confocal microscope. Image acquisition and analysis were performed using Olympus Fluoview FV10-ASW software.
Co-immunoprecipitation and western blotting
Plasmid combinations of interest (encoding HA- and Myc-tagged mouse CFAP53 and MNS1 proteins) were transfected into HEK293T cells in 6 cm dishes, with each plasmid at a dosage of 3 µg per dish, utilising Lipofectamine 2000 (Thermo Fisher Scientific). Following a 24-h incubation period, the transfected cells underwent lysis in 800 µl of RIPA buffer (Thermo Fisher Scientific) supplemented with complete Mini EDTA-free Protease Inhibitor Cocktail (Roche, 11836170001). The resulting cell lysates underwent brief sonication and centrifugation. An aliquot from the clear cell lysate was extracted and subjected to boiling in 1× SDS loading buffer as the input (total cell lysate). The remaining lysate was rotated overnight with 50 µl of Protein A-agarose beads (Roche) and 2 µg of mouse anti-HA (monoclonal, Santa Cruz Biotechnology, SC7392) or anti-Myc (Santa Cruz Biotechnology, SC289) antibodies. The beads were subjected to four washes in the immunoprecipitation (IP) buffer and then boiled in 50 µl of 1× SDS loading buffer (IP:HA). Both the total cell lysate (30 µl) and IP (30 µl) fractions were separated using SDS-PAGE gels, transferred to PVDF membranes, blocked in 3% bovine serum albumin, and probed with relevant primary antibodies [rabbit anti-HA (Santa Cruz Biotechnology, SC805, 1:3000) and rabbit anti-Myc (Santa Cruz Biotechnology, SC289, 1:3000)] and secondary antibodies [anti-mouse HRP conjugate (Promega, W4028) and anti-rabbit HRP conjugate (Promega, W4018) (both used at 1:3000)].
Live imaging with high-speed video microscopy
For high-speed video microscopy of cilia motility, zebrafish embryos were immobilised in 2% agarose (supplemented with 0.0175% Tricaine for 48 hpf embryos) and placed on 50 mm glass-bottom dishes. Ciliary motility was observed using a Zeiss 63× water-dipping objective mounted on an upright Zeiss AxioImager M2 microscope, which was equipped with an ORCA-Flash4.0 V2 C11440-22CU camera (Hamamatsu). Video processing and kymograph analyses were carried out using ImageJ 1.44d (NIH, USA). Mouse embryos and dissected tracheae were collected into DMEM-HEPES with 10% foetal bovine serum. Motility of cilia was examined at 25°C with a high-speed CMOS camera (HAS-500 M or HAS-U2, DITECT) at a frame rate of 100 frames/s. The pattern of ciliary motion was traced and analysed with ImageJ and Photoshop CC (Adobe).
TEM
E8.0 mouse embryos and adult tracheae were fixed with 2% paraformaldehyde and 2.5% glutaraldehyde in 0.1 M phosphate buffer pH 7.4, 4°C. The samples were then washed with 0.1 M phosphate buffer and exposed to 2% osmium tetroxide. Ultrathin (70 nm) sections were prepared with a diamond knife, mounted on 200-mesh copper grids, stained with 2% uranyl acetate and lead stain solution, and observed with a JEM-1400Plus microscope (JEOL). Preparation of samples and imaging were performed in the facility of Tokai Electron Microscopy, Inc., Japan.
Statistics
All statistical methods used are mentioned in the legends of relevant figures. Sample sizes were determined based on recessive Mendelian genetics for Mns1 and Cfap53 mutant alleles in zebrafish as well as mice.
Figure assembly
All figures were assembled using Adobe Illustrator CS4.
Acknowledgements
We thank H. L. Yeo for technical assistance with generation of the zebrafish cfap53 mutant strain, B. Durand for plasmids containing zebrafish odad1 and odad3 cDNAs, and A. Brown and M. Gui for critical reading and comments on the manuscript.
Footnotes
Author contributions
Conceptualization: H.H., S.R.; Methodology: H.L., W.K.T., Y.I., V.K., I.M., K.B.S., K.X.C., A.A.; Software: I.M., K.B.S.; Visualization: H.L., W.K.T., Y.I., H.H., S.R.; Validation: H.L., W.K.T., Y.I., V.K., I.M., K.X.C., A.A.; Formal analysis: H.L., W.K.T., Y.I., V.K., I.M., K.B.S., K.X.C., A.A.; Writing - original draft: S.R.; Writing - review and editing: S.C., H.H., K.B.S., S.R.; Supervision: S.C., H.H. S.R.; Funding acquisition: S.C., H.H., S.R.
Funding
This work was supported by funds from the Ministry of Education, Culture, Sports, Science and Technology (MEXT) of Japan (17H01435 to H.H.), the Council of Scientific and Industrial Research, India (to S.C.) and the Agency for Science, Technology and Research (A*STAR) of Singapore (to S.R.).
Data availability
All relevant data can be found within the article and its supplementary information.
Peer review history
The peer review history is available online at https://journals.biologists.com/dev/lookup/doi/10.1242/dev.202737.reviewer-comments.pdf
References
Competing interests
The authors declare no competing or financial interests.