Lrrcc1 and Ccdc61 are conserved effectors of multiciliated cell function

Multiciliated epithelia are composed of multiciliated cells (MCCs) harboring numerous motile cilia. Ciliary beating generates powerful strokes that are essential for a variety of physiological functions in animals (Meunier and Azimzadeh, 2016). In aquatic organisms of the Lophotrochozoan and Echinodermata phyla, coordinated MCC beating is required for locomotion, clearance and transport of particles, and for feeding of larvae. In vertebrates, MCCs produce external or internal fluid flows. In lungfish, the ciliated epidermis clears the animal of particles and settling organisms before hatching (Kemp, 1996). In amphibian embryos, several roles have been proposed for the ciliated epidermis – prevention of micro-organisms and debris from attaching to the epidermis, pre-hatching rotation and post-hatching gliding, respiratory gas exchange, movement of surface mucus films and transportation of chemical signals to the olfactory organs (Nokhbatolfoghahai et al., 2006). Which of those roles is carried out by MCCs of the Xenopus embryonic epidermis remains unclear, despite a recent wealth of mechanistic studies in this model (Boutin and Kodjabachian, 2019; Brooks and Wallingford, 2014). In mammals, among other functions, MCCs help circulation of the cerebrospinal fluid in the central nervous system, mucociliary clearance of pathogens and pollutants from airways, and transportation of gametes in genital tracts (Spassky and Meunier, 2017). Consequently, mutations in genes necessary for multiple cilia formation or beating cause familial syndromes characterized by severe chronic airway infections and an elevated risk of infertility (Boon et al., 2014; Wallmeier et al., 2014). A precise multiscale organization of ciliary beating is required to establish a robust and directed flow at the surface of ciliated epithelia. At the cellular scale, all cilia must beat in the same direction, and, at the tissue scale, the beating direction must be coordinated between neighboring cells. The planar cell polarity pathway plays a prominent role in the establishment and maintenance of cilia polarity, both at tissue and cellular scales (Boutin et al., 2014; Chien et al., 2015; Mitchell et al., 2009; Park et al., 2008; Walentek et al., 2017).

Motile cilia of MCCs are built via a multistep process (Boutin and Kodjabachian, 2019; Spassky and Meunier, 2017). First, numerous centrioles must be produced and subsequently released in the cytoplasm (Al Jord et al., 2014; Zhao et al., 2013). Next, centrioles migrate and dock at the apical surface, where they acquire a regular distribution and a coordinated orientation (Herawati et al., 2016; Park et al., 2008; Werner et al., 2011). Finally, ciliary axonemes extend from docked centrioles, and metachronal waves of ciliary beating are initiated and subsequently reinforced by mechanical feedback from the flow, which refines the coordination of cilia polarity (Guirao et al., 2010; Mitchell et al., 2007). During their journey towards the surface, centrioles mature into basal bodies (BBs) by acquiring basal foot (BF) and rootlet appendages, which localize asymmetrically and are essential for cilia polarization (Meunier and Azimzadeh, 2016). In Xenopus epidermal MCCs, two different types of rootlets attach to the proximal end of BBs. The most prominent rootlet has a fan shape and is localized opposite to the BF, which itself is positioned distally on the BB. The second rootlet is much longer and thinner and ‘dives’ into the cytoplasm (Zhang and Mitchell, 2015). These appendages confer an intrinsic polarity to BBs, which, in mature MCCs, reflects the direction of ciliary beating, with the BF pointing in the direction of the effective stroke.

Cilia formation and organization in MCCs relies on close interactions between BBs and cytoskeletal elements (Boutin and Kodjabachian, 2019; Meunier and Azimzadeh, 2016). The transport and docking of centrioles/BBs to the apical surface is dependent on acto-myosin-based mechanisms (Boisvieux-Ulrich et al., 1990; Epting et al., 2015; Kulkarni et al., 2018; Lemullois et al., 1988; Miyatake et al., 2015; Park et al., 2008). Once docked, neighboring BBs are linked by subapical actin filaments and apical microtubules (MTs) emanating from rootlets and BFs, respectively. The geometrical network hence made ensures regular BB spacing and coordinated orientation over the apical cell surface (Antoniades et al., 2014; Lemullois et al., 1988; Park et al., 2006; Werner et al., 2011; Yasunaga et al., 2015). Chemical interference with actin or MT networks leads to BB disorganization and impaired ciliary function (Herawati et al., 2016; Werner et al., 2011). Reciprocally, depletion of specific centriolar components, which prevents the formation of appendages and/or precludes BB–cytoskeleton interactions, alters BB organization and ciliary function (Antoniades et al., 2014; Bustamante-Marin et al., 2019; Clare et al., 2014; Herawati et al., 2016; Kulkarni et al., 2018; Kunimoto et al., 2012; Turk et al., 2015; Walentek et al., 2016).

Leucine rich repeat coiled-coil domain containing 1 (Lrrcc1) and coiled-coil domain-containing protein 61 (Ccdc61) are structural proteins, which are conserved from Chlamydomonas to human, and involved in centriole appendage biogenesis and function both at the centrosome and at ciliary BBs (Adams et al., 1985; Bärenz et al., 2018; Basquin et al., 2019; Bengueddach et al., 2017; Hoops et al., 1984; Muto et al., 2008; Ochi et al., 2020; Pizon et al., 2020; Silflow et al., 2001; Wright et al., 1983; Gaudin et al., 2021). In MCCs of the planarian epidermis, Lrrcc1 (where it is also known as Vfl1) and Ccdc61 (known as Vfl3) depletion causes structural BB defects, thus perturbing cilia orientation, and altering the direction of locomotion (Basquin et al., 2019). The functional importance of Lrrcc1 and Ccdc61 in vertebrate MCCs remains unknown, although Ccdc61 was recently reported to associate with BBs in Xenopus MCCs (Ochi et al., 2020). Here, we used the Xenopus laevis ciliated epidermis as a model to shed light on this issue. We decided to comparatively study Lrrcc1 and Ccdc61, based on their shared biological functions in Chlamydomonas and Schmidtea (Adams et al., 1985; Basquin et al., 2019; Hoops et al., 1984; Wright et al., 1983), and their reported physical interaction in an unbiased human proteomic screen (Hein et al., 2015). In Xenopus, MCCs are specified deeply, in the inner cellular layer of the epidermis, before intercalating at regular intervals into the outer cellular layer (Chuyen et al., 2021; Collins et al., 2021, 2020; Deblandre et al., 1999; Stubbs et al., 2006; Werner et al., 2014). BB synthesis is initiated while MCCs are still in the inner epidermal layer (Klos Dehring et al., 2013; Revinski et al., 2018); BB docking, distribution, orientation and ciliogenesis are completed when MCCs have radially intercalated and expanded their apical surface (Chung et al., 2014; Kulkarni et al., 2021; Sedzinski et al., 2016).

We report here that Lrrcc1 and Ccdc61 are both associated with Xenopus MCC centrioles, with a preferential localization near the basis of ciliary rootlets in mature MCCs. We found that knocking down either gene impacts ciliated epithelium biogenesis at multiple scales. At the organelle scale, Ccdc61 is required for rootlet association of Pericentrin (Pcnt), which we characterize as a novel marker of this appendage in Xenopus BBs. At the cellular scale, we show that Lrrcc1 and Ccdc61 are required for proper organization of BBs. In addition, MCCs depleted for either gene present important defects in apical cytoskeleton organization. Finally, at the scale of the embryo, Lrrcc1 and Ccdc61 knockdown revealed their importance for generation of superficial fluid flows and the resistance to opportunistic pathogens. This study bridges multiple scales of analysis to reveal how intracellular disorganization of MCCs can impair the physiology of the whole organism.

Single-cell RNA sequencing (scRNA-seq) of Xenopus tropicalis embryos previously revealed that lrrcc1 and ccdc61 are specifically expressed in ciliated epidermal cells from gastrula stages onwards (Briggs et al., 2018), and both genes were found to be activated by the complex of Multicilin with E2F4 or E2F5 complex, which is known to be necessary and sufficient for vertebrate MCC differentiation (Ma et al., 2014). We used whole mount in situ hybridization to analyze the localization of lrrcc1 and ccdc61 transcripts in Xenopus laevis embryos. At early tailbud stage 20, both genes displayed a ‘salt and pepper’ pattern typical of epidermal MCCs (Fig. S1A). Double fluorescence in situ hybridization confirmed expression in MCCs marked by α-tubulin (Fig. S1B).

Next, we used fluorescent immunostaining to analyze the distribution, by confocal microscopy, of the endogenous Lrrcc1 and Ccdc61 proteins at different time points during MCC differentiation. At stage 18, when released centrioles start migrating towards the apical surface, both proteins appeared closely associated with individual neo-centrioles marked by Centrin (Fig. 1A,G). At stage 31, both Lrrcc1 and Ccdc61 were found associated with mature BBs docked at the apical surface (Fig. 1B,H). The Lrrcc1 signal was systematically located on one side of Centrin-positive BBs. In a magnified orthogonal view, the signal could be seen to flank the BB and extend towards the cytoplasm (Fig. 1B). The Ccdc61 signal overlapped considerably more with Centrin in both top and lateral views (Fig. 1H). To assess with more precision Lrrcc1 and Ccdc61 localization, we analyzed their distribution in 3D, relative to Centrin and γ-Tubulin, which is known to localize to the BF cap in MCCs (Clare et al., 2014; Hagiwara et al., 2000). On a top view, Centrin appeared as a single dot (Fig. 1E,K). Unexpectedly, two pools of γ-Tubulin were detected adjacent to the BB. The first pool appeared as a dot with strong intensity juxtaposed to the BB core. A lateral view revealed that this dot was located at a depth similar to that of the BB (Fig. 1E,K), thus corresponding to the γ-Tubulin previously described at the BF. In contrast, the second pool of γ-Tubulin displayed a less intense signal, localized opposite to the BF with respect to the BB, and extended into the cytoplasm (Fig. 1E,K), in a position compatible with the fan-shaped rootlet described in Xenopus MCCs (Zhang and Mitchell, 2015). To further assess this possibility, we applied Centrin and γ-Tubulin immunostaining on embryos injected with Clamp–GFP, an accepted marker of the fan-shaped rootlet in Xenopus (Park et al., 2008). This analysis confirmed that the weaker pool of γ-Tubulin is at the rootlet (Fig. S1C). 3D analysis revealed that the Lrrcc1 signal was localized adjacent to the rootlet pool of γ-Tubulin (Fig. 1E). A similar localization was observed in MCCs expressing an Lrrcc1–GFP fusion construct (Fig. S1D,E). Triple staining with Centrin and Clamp–GFP revealed that the Lrrcc1 signal was concentrated at the junction between the BB and the rootlet (Fig. 1F). We note, however, that the Lrrcc1 signal detected below Centrin in uninjected MCCs (Fig. 1E), was strongly diminished in Clamp–GFP-injected MCCs (Fig. 1F). This could reflect displacement of Lrrcc1 by Clamp–GFP, or reduced Lrrcc1 antibody accessibility. Triple staining with Centrin and Clamp–GFP revealed that the Ccdc61 signal was localized at the junction between the BB and the rootlet (Fig. 1L). Consistent with this, a Ccdc61–GFP fusion construct localized opposite to the BF pool of γ-Tubulin (Fig. S1F,G). A similar localization was recently reported for a Ccdc61–RFP fusion in Xenopus epidermal MCCs (Ochi et al., 2020).

We conclude that Lrrcc1 and Ccdc61 preferentially localize at the junction between the BB and the rootlets in mature epidermal MCCs of Xenopus.

Taken together, these data suggest that Lrrcc1 and Ccdc61 are associated with epidermal MCC centrioles from the time of their release in the cytoplasm through the phases of docking, BB maturation, ciliary growth and maintenance.

The subcellular distribution of Lrrcc1 and Ccdc61 during MCC differentiation suggested that they could be involved in BB apical docking as well as distribution and orientation at the apical surface. To investigate this, we knocked down lrrcc1 and ccdc61 through injection in the presumptive epidermis of two independent morpholino antisense oligonucleotides (MOs) designed to block translation (MO-ATG) or splicing (MO-Spl) (Fig. S2A). The capacity of translation MOs to inhibit protein synthesis was verified both by western blotting and immunofluorescence of endogenous or GFP-tagged proteins (Fig. S2B–G). We performed immunostaining at stage 31 to analyze BB organization in morphant MCCs. As previously described (Werner et al., 2011), BBs displayed a stereotypical organization in mature control MCCs. They were all docked, quite evenly distributed at the apical surface and oriented in a coordinated manner (Fig. 2A,C,E). In contrast, Lrrcc1 and Ccdc61 knockdown drastically impaired centriole organization, and morphant cells could be classified in two phenotypic categories. The first phenotypic class was characterized by clusters of centrioles stuck in the upper half of the cytoplasm (Fig. 2A; Fig. S3A). To quantify this defect, we analyzed the apico-basal (A–B) distribution of centrioles along the Z-axis of confocal stack acquisitions. In control cells, most centrioles localized within the first 1.6 µm below the apical cell surface (Fig. 2B). In contrast, in Lrrcc1-ATG and Ccdc61-ATG morphant cells most centrioles localized deeper, between 1.6 µm and 5.6 µm from the surface (Fig. 2B). Similar results were obtained with Lrrcc1-Spl and Ccdc61-Spl MOs (Fig. S3B). In the second phenotypic class, most centrioles properly localized at the apical surface, but displayed irregular spacing and a randomized orientation (Fig. 2C,E; Fig. S3C,E). To quantify BB spacing defects, we applied Delaunay triangulation (Scarpa et al., 2015) between the centroids of docked centrioles and measured the area of the obtained triangles. In the control situation, the even distribution of BBs resulted in many triangles of similar area that tightly distributed around the median (Fig. 2C,D; Fig. S3C,D). In contrast, in morphant cells, variable distances between BBs resulted in a broader distribution of triangles around the median (Fig. 2C,D; Fig. S3C,D). We revealed BB orientation by immunostaining the BB core (Centrin) and the offset BF (intense γ-Tubulin spot), which allowed us to automatically extract orientation vectors and plot their circular distribution. In control cells, vector angles tightly distributed around the mean, indicating that BBs oriented in the same global direction. In Lrrcc1 and Ccdc61 morphant cells, the intense γ-Tubulin signal was largely preserved, suggesting that the integrity of the BF was maintained. However, vector angles were widely distributed around the mean, indicating a randomization of BB orientation (Fig. 2E; Fig. S3E). Accordingly, circular standard deviation values were significantly higher in morphants, as compared to control (Fig. 2F; Fig. S3F). The Rayleigh statistical test also revealed a higher percentage of morphant MCCs in which no significant mean vector could be defined, as compared to control (Fig. 2G; Fig. S3G).

In both Lrrcc1 and Ccdc61 morphant MCCs, cilia could be observed, suggesting that those proteins are not required for ciliary growth per se (Fig. S4). Importantly, for both genes and both types of MOs, docking, spacing and orientation phenotypes were rescued by co-injection of lrrcc1 or ccdc61 mRNA constructs lacking (lrrcc1), or silently mutated (ccdc61) on the MO-binding sequence (Fig. 2B,D,F; Figs S3B,D,F and S5).

Taken together, these data show that Lrrcc1 and Ccdc61 are required for correct apical migration and/or docking of BBs, as well as for their proper planar distribution and orientation at the apical surface.

Our localization and functional data suggested that Lrrcc1 and Ccdc61 participate in the formation of rootlet appendages in Xenopus MCCs. To analyze this possibility, we performed transmission electron microscopy (TEM) experiments on stage 31 MCCs from control and morphant embryos. The presence of centriole docking defects was used as a way to ascertain the morphant character of scored MCCs (Fig. 3B). Both fan-shaped and thin rootlets could be observed in similar proportions in Lrrcc1 and Ccdc61 morphant cells, as compared to control (Fig. 3A,C). Whenever visible, rootlets did not present obvious structural or positioning defects.

Next, we looked for potential molecular defects of ciliary rootlets in morphant MCCs. In the course of an independent study about its role in MCC biogenesis (unpublished), we produced a polyclonal antibody against Xenopus Pcnt (Fig. S6A,B), which allowed us to uncover its association to ciliary rootlets. In stage 31 control cells, Pcnt immunostaining presented a dotted pattern, and was associated with Centrin and γ-Tubulin at the apical surface (Fig. 3D). 3D analysis of the relative distribution of these proteins in individual BBs revealed that Pcnt was located opposite to the strong BF-associated dot of γ-Tubulin, in a plane below Centrin. It was present as one or two dots emerging from the BB and extending towards the cytoplasm (Fig. 3D′). Pcnt association to rootlets was further confirmed by its colocalization with Clamp–GFP (Fig. 3E,E′). This series of tests revealed that in Xenopus epidermal MCCs, Pcnt specifically localizes at rootlets.

Next, we analyzed Pcnt signal in Lrrcc1 and Ccdc61 morphant embryos. To avoid immunostaining variability between different embryos, we compared Pcnt signal intensity within mosaic embryos. For Lrrcc1, no differences were observed between non-injected and morphant cells, which, however, clearly displayed randomized BB polarity (Fig. 3F,G). In contrast, a marked decrease of Pcnt signal was observed in Ccdc61 morphant cells, as compared to non-injected cells from the same embryos (Fig. 3F,G).

Taken together, these experiments suggest that Lrrcc1 and Ccdc61 are not essential for building rootlet appendages in Xenopus MCCs. However, Ccdc61 is required for association of the protein Pcnt to the rootlet, which could affect rootlet function and MCC organization.

BB organizational defects observed in Ccdc61 and Lrrcc1 morphant cells might be linked to a defective apical cytoskeleton. To test this idea, we first analyzed F-actin networks in MCCs at stage 31. In control cells, we observed the stereotypical organization in apical and subapical networks that was previously described for mature MCCs (Werner et al., 2011). The apical actin network was organized like a grid surrounding each centriole, and the subapical network located just below was composed of short actin fibers that are known to connect neighboring BBs via their rootlets (Fig. 4A,A′). Both Lrrcc1 and Ccdc61 knockdown caused a global decrease of F-actin staining (Fig. 4D,E). When looking at individualized centrioles, a strong reduction of both apical and subapical F-actin was observed (Fig. 4B′,C′). From these results, we conclude that Lrrcc1 and Ccdc61 are, either directly or indirectly, involved in the assembly of the apical filamentous actin networks.

Next, we analyzed the apical MT network in mature MCCs. At stage 31, anti-α-Tubulin antibodies mainly revealed cilia, precluding the analysis of intracellular MT networks in mature MCCs (Fig. S7). To circumvent this limitation, we adopted a deciliation strategy to deplete cilia-associated signals and visualize intracellular MTs (Fig. S7). At stage 31, control cells were characterized by a highly organized apical MT network connecting BBs together (Fig. 5A,A′), similar to what has been reported by another method (Werner et al., 2011). In Lrrcc1- and Ccdc61-depleted MCCs, intense α-Tubulin signals were found associated with clustered centrioles (Fig. 5B,B′,D,D′). When looking at individualized BBs, MTs were clearly visible but the size and spatial organization of filaments appeared heterogeneous, as compared to control (Fig. 5C,C′,E,E′). These results suggest that Lrrcc1 and Ccdc61 are not necessary for apical MT network polymerization per se. Thus, the apparent MT network disorganization may be secondary to the loss of BB polarity.

Finally, we analyzed the apical intermediate filament (IF) network using the anti-cytokeratin C-11 antibody. At stage 18, we did not detect IFs inside or at the MCC surface, suggesting that they are not involved in centriole apical migration (Fig. 6A). At stage 31, IFs were organized into a dense grid surrounding each BB (Fig. 6B,B′), similar to what has been described in tracheal MCCs (Tateishi et al., 2017). Strikingly, IF organization was drastically affected in Lrrcc1 and Ccdc61 morphant cells. Overall, the lattice appeared much less dense, and the annular organization of IF around BBs was lost (Fig. 6C,C′,D,D′). This analysis suggests that Lrrcc1 and Ccdc61 are, directly or indirectly, involved in the establishment of the tight IF network in which BBs are embedded.

Taken together, these analyses reveal that Lrrcc1 and Ccdc61 depletion has profound impacts on MCC apical cytoskeleton organization.

The perturbed organization of BBs in morphant MCCs is expected to disturb the function of associated cilia, thereby affecting the production of fluid flow at the surface of the embryo. To address this issue, we first analyzed ciliary beating frequency by high-speed video recording. As previously described, control MCCs performed synchronized and large amplitude effective and recovery strokes, characterized by the ability of cilia to extensively bend (Fig. 7A; Movies 1 and 2) (Werner et al., 2011). In Lrrcc1 and Ccdc61 morphant embryos, we observed three levels of beating defaults: (1) low-amplitude, uncoordinated and disoriented beating, causing occasional collisions between cilia; (2) cilia performing only small vibrations; (3) extreme cases with totally immobile cilia (Fig. 7A; Movies 1 and 2). Accordingly, the beating frequency, which was ∼20 Hz for the control condition, decreased to lower values in a large majority of morphant MCCs (Fig. 7A). Next, we analyzed cilia-generated flow in the surrounding liquid, by live recording of visible dyed microspheres dispersed in the fluid, along the flanks of embryos at stage 31. In control condition, microspheres were moved by a robust flow and traveled the entire length of the embryo in ∼12 s. In contrast, the flow was severely slowed down in Lrrcc1 and Ccdc61 morphant conditions, and the microspheres rarely reached the middle of the embryo after 12 s (Fig. 7B,C and Movie 3).

To address the impact of impaired ciliary flow on the physiology of morphant individuals, we analyzed their susceptibility to pathogen infection. Embryos were incubated for 72 h with the opportunistic bacteria Aeromonas hydrophila (Dubaissi et al., 2018), and their survival rate was recorded. Non-injected and GFP-injected embryos were used as controls. For both control conditions, an almost complete survival rate was observed independently of the presence or not of bacteria (Fig. 7D). In absence of bacteria, Lrrcc1 and Ccdc61 morphant embryos displayed a slightly decreased survival rate as compared to control embryos. However, the presence of pathogenic bacteria strongly impacted the survival rate of morphants, starting from 24 h of incubation. We confirmed by immunostaining that centrioles and cilia were disorganized in morphant embryos from the same experimental series (Fig. S8). This assay suggests that Lrrcc1 and Ccdc61 inactivation leads to a higher susceptibility of embryos to pathogen infection, likely due to reduced cilia-powered clearance.

In this study, we report a role for Lrrcc1 and Ccdc61 in Xenopus epidermal MCC differentiation and function, extending their evolutionarily conserved importance in ciliated cells to vertebrates. At the cellular scale, both Lrrcc1 and Ccdc61 were necessary for the migration and/or docking, spacing and polarization of BBs at the cell surface and for apical cytoskeleton organization. At a larger scale, both factors were necessary for cilia-powered superficial flow to help survival of the organism when exposed to environmental pathogens.

Lrrcc1 and Ccdc61 proteins both contain coiled-coil domains, which are one of the most common structural motifs mediating protein–protein interactions. Searching in the IntAct protein interaction database reveals that Lrrcc1 and Ccdc61 physically interact with multiple known factors related to centrosomes, cilia and MT organization or polymerization, among which 30% are actually shared between the two proteins (Table S1) (Hein et al., 2015). Moreover, 20/26 (Lrrcc1) and 12/23 (Ccdc61) of those interactors were found to be expressed in MCCs of Xenopus tropicalis, as revealed by scRNA-seq (Table S1) (Briggs et al., 2018). Together with their reported interaction, this information is consistent with Lrrcc1 and Ccdc61 showing a shared localization and knockdown leading to the same phenotypes, and suggest that these two factors belong to a common molecular network important for BB function in MCCs.

We show that Ccdc61 and Lrrcc1 associate to BBs early after their production and maintain this association in fully differentiated MCCs. The early association to neo-synthesized centrioles is compatible with a role in apical migration, which appears to be incomplete in Lrrcc1- and Ccdc61-deficient mature MCCs. Alternatively, apical migration may be unaffected, and defective docking may secondarily cause BBs to ‘dive’ back into the cytoplasm.

The distribution of Lrrcc1 and Ccdc61 at BBs is consistent with their observed requirement for correct MCC organization and function. This role is shared with the Vfl1 (Lrrcc1) and Vfl3 (Ccdc61) orthologs, which are required to organize flagella/cilia in Chlamydomonas, Paramecium and planarians (Adams et al., 1985; Basquin et al., 2019; Bengueddach et al., 2017; Hoops et al., 1984; Wright et al., 1983). In all these species, these two factors are required for the proper construction of centriolar appendages. Based on our data and this literature, we propose that Lrrcc1 and Ccdc61 may be involved in rootlet appendage formation and/or function in Xenopus MCCs. In contrast to studies in the other species, however, we did not detect major rootlet structural defects upon Lrrcc1 or Ccdc61 depletion, suggesting the existence of redundant mechanisms to build up Xenopus BB appendages.

From their place of birth to their final position at the apical surface, the relocation of centrioles/BBs is intimately linked to cytoskeletal networks (Boisvieux-Ulrich et al., 1990; Herawati et al., 2016; Lemullois et al., 1988; Werner et al., 2011). The organization of these networks evolves during MCC maturation, allowing first to direct BBs towards the apical surface and then to orient and space them. By using chemical cytoskeleton inhibitors, it has been possible to alter migration, orientation and dispersion of BBs, thus attributing specific functions to actin filaments and MTs (Boisvieux-Ulrich et al., 1990; Werner et al., 2011). Our data revealed that establishment of proper actin filamentous networks require the presence of both Lrrcc1 and Ccdc61at BBs. Our data, however, do not help to resolve how these proteins participate in actin network assembly. They could, for instance, be necessary for BBs to nucleate actin filaments in a RhoA-dependent manner (Chevalier et al., 2015; Pan et al., 2007; Park et al., 2008). Alternatively, they could be involved in the formation, maintenance or activity of ciliary adhesions, which link BBs and rootlets with actin filaments (Antoniades et al., 2014). Future studies should address these and other possibilities to help elucidating the precise implication of Lrrcc1 and Ccdc61 in MCC actin network formation. In contrast to actin filaments and MTs, the contribution of IFs to the organization of MCCs remains unknown. Our analysis did not reveal the presence of IFs at the stage of centriole apical migration, ruling out their involvement at this step. In mature MCCs, a prominent IF network adopts an annular shape around BBs, similar to the apical F-actin network, but which appears to form more basally, extending below the BB level, where rootlets are found (Figs 4A′ and 6B′) (Sandoz et al., 1988). This IF network collapsed in MCCs deprived of Lrrcc1 and Ccdc61, suggesting that IFs could interact with the base of BBs and/or ciliary rootlets through molecular complexes dependent on these two structural proteins. This first rudimentary analysis makes the Xenopus epidermis an attractive paradigm to address the specific role of IFs in MCC organization and function. In particular, it would be interesting to evaluate the possible link between actin and IF networks.

In contrast to the BF, which functions as an MT-organizing center (Clare et al., 2014; Kunimoto et al., 2012), the role of the rootlet appendage is less clear. Among proposed functions, it is generally believed to serve as an anchor, allowing BBs to maintain their position at the apical surface, and resist mechanical forces generated by ciliary beating (Antoniades et al., 2014; Bustamante-Marin et al., 2019; Yang et al., 2005; Yang and Li, 2005). In line with this, rootlets are scaffolds for molecular interactions, among which ciliary adhesion complexes have been found to organize the short actin filaments that cross-link BBs together and help beating synchronization (Antoniades et al., 2014; Walentek et al., 2016; Werner et al., 2011). Here, we observed that Ccdc61, Pcnt and γ-Tubulin associate to rootlets in mature epidermal MCCs of Xenopus. Ccdc61 was recently shown to be a paralog of the scaffolding protein Sas6, known to establish the ninefold rotational symmetry of MTs during centriole duplication (Ochi et al., 2020). Interestingly, an evolutionarily conserved interaction between Sas6 and Pcnt has recently been reported (Ito et al., 2019). Furthermore, Pcnt is known for its ability to recruit γ-Tubulin to assemble a macro-molecular complex allowing MT nucleation at the centrosome, and at mitotic spindles (Woodruff et al., 2014). Based on this information and our own observations, it is tempting to propose that Ccdc61 may help Pcnt recruitment/stabilization at the rootlet, which in turn would favor γ-Tubulin recruitment and MT nucleation. Alternatively, the reported interaction of Ccdc61 with MTs (Ochi et al., 2020) might be important to link pillar MTs emanating from the BF (Clare et al., 2014) with nearby rootlets, which could further strengthen the mechanical coupling between adjacent BBs, to optimize their coordinated orientation and synchronized beating. In support of this hypothesis, it is relevant to recall that Clamp (also known as Spef1) is also a MT-binding factor (Dougherty et al., 2005; Kim et al., 2018; Werner et al., 2014). Future super-resolution and EM studies should investigate the possible link between MTs and rootlets.

Unexpectedly, we found that cilia in Lrrcc1 and Ccdc61 morphant MCCs often beat very poorly. Proper ciliary beating entails correct axonemal structure and delivery of dynein motors to maintain ciliary motion (Huizar et al., 2018; Satir et al., 2014). Although we have not investigated these features, it is possible that the lack of Lrrcc1 and Ccdc61 compromises the capacity of rootlets to help trafficking towards the axoneme of essential structural or motor effectors (Gray et al., 2009; Mohan et al., 2013; Park et al., 2008; Yang and Li, 2005).

From the nanoscopic organization of organelles to the cellular function and physiology of the organism, all scales are coupled. The emergence of the locomotor function in planarians is a striking example of such coupling in multiciliated epithelia. In this model, polarization of BBs relies on their chiral construction, which mobilizes Lrrcc1 and Ccdc61. This polarization allows the establishment of bilateral symmetry of the ventral multiciliated epidermis, which in turn governs the orientation of the worm movement (Basquin et al., 2019). In addition, retro-control loops exist between the different scales. For instance, in mouse ependymal MCCs, the establishment of a dense actin network that confers stability to BBs and cilia is dependent on active ciliary beating (Mahuzier et al., 2018). Thus, understanding such highly integrated systems implies analyzing multiple scales, as well as their coupling and feedback mechanisms. The present multi-scale study correlates the absence of structural proteins at BB appendages to MCC disorganization, defective flow production and impaired resistance to pathogens. Despite a wealth of studies on the Xenopus ciliated embryonic skin, it was unknown whether cilia-powered flow could limit infections by environmental pathogens. In that respect, the function of the Xenopus mucociliary epidermis is similar to that of the mucociliary epithelium that ensures the clearance of incoming pathogens in mammalian airways. In airways, cilia beating helps to propel a viscous layer of mucus trapping foreign particles, which lies on top of an aqueous periciliary layer. In contrast, in the Xenopus skin, the mucus layer is ∼6 µm thick and sits right on top of the epithelium (Dubaissi et al., 2018), so that the 15-µm-long cilia actually beat in water to prevent attachment of micro-organisms. Thus, unlike in airways, the mucus on the frog embryonic skin may not be propelled as a coherent layer, which would explain the much lower density of MCCs necessary in this system. Although additional functions, such as oxygenation of the skin, are not ruled out, our proposed pathogen clearance function is consistent with the resorption of MCCs in pre-metamorphic tadpoles (Tasca et al., 2020), at a stage when innate immunity is in place (Robert and Ohta, 2009).

All experiments were performed following the Directive 2010/63/EU of the European parliament and of the council of 22 September 2010 on the protection of animals used for scientific purposes and approved by the “Direction départementale de la Protection des Populations, Pôle Alimentation, Santé Animale, Environnement, des Bouches du Rhône” (agreement number F1305521).

cDNA fragments from Xenopus laevis lrrcc1.L (Entrez gene 431936), and ccdc61.L (Entrez gene 734655) were amplified from commercial cDNA (Horizon discovery) by PCR using the following primers: lrrcc1 forward, 5′-GCGAACGGACACAGACAGTA-3′; lrrcc1 reverse, 5′-GAATTCCATGGTAGTCAGCTCCTGC-3′; ccdc61 forward, 5′-GCGGCCGCAAGTGGAGGATGCTGTGACC-3′; and ccdc61 reverse, 5′-GAATTCACGGATGAACTGCGTCTCTG-3′.

PCR products were cloned in pBlueScript KS+ vector (Agilent) and digoxigenin-labeled probes were generated from linearized plasmids using RNA-labeling mix (Roche). Whole-mount chromogenic in situ hybridization was performed as described previously (Marchal et al., 2009) using 40 ng of digoxygenin-labeled probe. Pictures were taken with the stereomicroscope Leica MZ125 coupled to NIKON digital Sight DS-Fi1 camera.

To generate plasmids for RNA synthesis and micro-injection, the ORFs of lrrcc1 and ccdc61 were amplified by PCR using the following primers: lrrcc1-forward, 5′-TCTTTTTGCAGGATCACAATGGCAGGCACGGACCCACGAA-3′; lrrcc1-reverse, 5′-CTTTACTCATTCTAGAAAATTCTTTTTGGATGTCACTTAG-3′; ccdc61-forward, 5′-TCTTTTTGCAGGATCACAATGGAGGATACAGAGTTTGCT-3′; and ccdc61-reverse, 5′-CTTTACTCATTCTAGACTGCATCAGTAAGTACCCGCTGGCT-3′.

PCR products were subcloned in frame with GFP sequence in 3′ into pCS2+-GFP (lab stock) vector using an In-Fusion^® HD Cloning Kit (Takara Bio USA, Inc.). For rescue experiments, silent mutations were introduced by PCR in the original Ccdc61 sequence (3′- GAGGATACAGAGTTTGCTGAAG-5′) to generate a Ccdc61-GFP construct (named MOresCcdc61-GFP) with a sequence resistant to MO ATG (3′-GAAGACACGGAATTCGCTGAA-5′). pCS2+-mRFP (lab stock) was used to generate an injection reporter.

The cDNA fragment coding for amino acids 3340 to 3643 of the full-length Xenopus laevis Pcnt was amplified by PCR from a partial cDNA clone (IMAGE 5156155, Source Bioscience). PCR products were cloned by Gateway^® recombination into pGEX6P3 (GE Healthcare Life Sciences) and pEGFP-C1 (Clontech) to produce GST fusion proteins and express GFP-tagged proteins, respectively. The Centrin-RFP plasmid was a kind gift from John B. Wallingford (Dept. of Molecular Biosciences, University of Texas at Austin, USA).

Capped mRNAs were synthetized from linearized vectors using the SP6 mMESSAGE mMACHINE^® Kit (Ambion Life Technologies) and purified with the MEGAclear™ Kit (Ambion Life Technologies).

Two independent morpholino antisense oligonucleotides were designed against lrrcc1 and ccdc61 (GeneTools, LLC). lrrcc1-ATG-MO, 5′-GTGCCTGCCATTCTCCCGCAACAAA-3′; lrrcc1-Spl-MO, 5′-ACTGAAGCCATGCTGCTTACCTGGA-3′; ccdc61-ATG-MO, 5′-CTTCAGCAAACTCTGTATCCTCCAT-3′; ccdc61-Spl-MO, 5′-TGTCTCCCACTTCTACTCACATTGA-3′.

Eggs obtained from wild-type Xenopus laevis females of 2 to 5 years of age (NASCO, USA) were fertilized in vitro, dejellied and cultured as described previously (Marchal et al., 2009). 20–30 ng of MO was injected (alone or with 200–500 pg of mRFP tracer) in one or two animal-ventral blastomere (presumptive epidermis) at different stages depending on the experiment (Table S2). Working amounts of MOs were first calibrated to optimize embryo survival. Validation of MOs efficacy in depleting the target protein was confirmed based on Lrrcc1 antibody and Ccdc61–GFP signal disappearance. For rescue experiments, a sequential injection strategy was adopted to obtain mosaic embryos containing differentially marked morphant and rescued MCCs. At the four-cell stage, lrrcc1 and ccdc61-ATG-MOs plus Centrin–RFP, or lrrcc1 and ccdc61-Spl-MOs plus Centrin–RFP was injected. At the 16-cell stage, the same embryos were injected with Lrrcc1–GFP or MOresCcdc61-GFP mRNAs.

After injection, embryos were incubated at different temperatures between 13°, 18° and 23°C in 0.1× MBS (Fluka D0638-16) until they reached the desired developmental stage. Whole-embryos were fixed in different conditions (Table S3) and stored in 100% methanol at −20°C in preparation for immunofluorescent staining.

Stage 31 embryos were incubated for 2 h in a 35-mm Petri dish containing 2 ml dibucaine hydrochloride (200 µM in 0.1× MBS; Fluka D0638-16). After a rapid wash in 0.1× MBS, embryos were immediately fixed by incubation in 100% methanol at −20°C for 2 days before being further processed for immunofluorescence staining.

After a sequential re-hydration in solutions with decreasing methanol concentrations, embryos were incubated in blocking solution [3% bovine serum albumin (BSA) in 1× PBS] for 1 h at room temperature (RT) and subsequently in primary antibodies diluted in 3% BSA overnight at 4°C (Table S3). After washing in 1× PBS embryos were incubated with fluorescently labeled secondary antibodies diluted in 3% BSA for 1 h at RT (Table S4). After washing, embryos were mounted in Mowiol (Sigma-Aldrich) between slide and coverslip.

Confocal pictures were acquired using ZEISS LSM 780 right standing AxioImager Z2 and ZEISS LSM 880 reverse standing AxioObserver 7 equipped with 20× and 63× oil objectives. Two- or three-colors confocal Z-series images (Z-slices interval between 0.3 and 0.8 nm) were acquired using sequential laser excitation. When necessary, images were converted into single plane maximum intensity projection (MIP) images and edited using Image J 2.0.0 software. ClearVolume Plugin for ImageJ (Royer et al., 2015) was used for 3D visualization of basal bodies.

F-actin and Pcnt signals were quantified by measuring the mean pixel intensity per MCC with ImageJ.

BB docking was quantified manually by counting the number of Z-slices containing Centrin- or Centrin–RFP-positive centrioles along the apical-–asal axis for each cell. These values were transformed (in µm) based on the interval size between slices set during confocal acquisition.

BB spacing analysis was performed from Centrin or Centrin–RFP immunostaining images using custom made Matlab scripts (available upon request). After manual segmentation of individual cells, the script (1) automatically detects, and segment individual BBs within each cell. Only apically docked and isolated BBs are considered, omitting those located below the apical cell membrane or in dense clusters; (2) determine the centroid of each BB; (3) use XY coordinates of all the centroids to build triangles between the nearest neighbors with the Delaunay Triangulation Matlab script. Once the triangulation is obtained, areas of the triangles are measured in pixel² and transformed in µm² based on the pixel size of each acquisition.

BB orientation was analyzed using a home-made ImageJ script (designed by R. Flores-Flores). The script (1) automatically detects individual MCCs in the field of view; (2) automatically detects individual BBs in each MCC; (3) traces a vector from Centrin to γ-Tubulin spots (a threshold is applied for γ-Tubulin channel in order to detect only the BF associated spot, and not the rootlet spots) for each BB and calculate its angle with the vertical axis. The output of the script is a list of angles that were plotted using the Oriana software (Version 4.02, Kovach Computing Services) to obtain a graphical representation of their distribution in 25.7° bins, with 95% confidence intervals. Finally, circular statistical analysis were performed (CSD, Rayleigh's uniformity test).

Embryos were processed for electron microscopy as previously described (Revinski et al., 2018) and cut transversely at midbody level (80 nm/slice) with a Leica Ultracut UC7 (Leica, Germany). Images were acquired using a Tecnai G2 (Thermofisher, USA) microscope equipped with a Veleta camera (Olympus, Japan).

Simian COS-1 cells (a gift from Françoise Birg, CRCM, Marseille, France) were grown in DMEM supplemented with 10% heat inactivated FCS and transfected with Fugene HD (Roche Applied Science) according to the manufacturer's protocol. Transfected or control cells were washed in PBS and lysed in 50 mM Tris-HCl pH 7.5, 150 mM NaCl, 1 mM EDTA, containing 1% NP-40 and 0.25% sodium deoxycholate (modified RIPA) plus a Complete protease inhibitor cocktail (Roche Applied Science) on ice. Cell extracts separated on polyacrylamide gels were transferred onto Optitran membrane (Whatman), followed by incubation with rabbit anti-GFP (1:5000, Abcam, ab290), or rabbit anti-Pcnt antibody (made in-house; 1 µg/ml) produced by immunization with recombinant portion of Xenopus laevis Pcnt (XP_018091513.1, residues 3340–3643), and horseradish peroxidase-conjugated secondary antibody (Jackson Immunoresearch Laboratories, 711-035-152). Signal obtained from enhanced chemiluminescence (Western Lightning ECL Pro, Perkin Elmer) was detected with an MyECL Imager (ThermoFisher Scientific).

For MO validation by western blotting, animal caps from uninjected embryos as a control, or animal caps injected with Lrrcc1–GFP with or without MO, or Ccdc61–GFP with or without MO, were obtained by manual dissection from stage 10 (n=50 embryos/condition) in 1× MBS and kept in 0.5× MBS until matched control embryos reached stage 25. Then they were immediately lysed in 200 µl of RIPA buffer (50 mM Tris-HCl pH 7.5, 150 mM NaCl, 1% NP40, 0.1% SDS and 0.5% sodium deoxycholate) containing a protease inhibitors tablet (Pierce). Total protein concentration was determined by Bradford assay (Invitrogen) and samples were prepared in LDS sample buffer (Invitrogen) containing a reducing agent. Samples in LDS buffer were denatured for 5 min at 100°C and 120 µg of proteins were loaded and separated on 4–20% SDS-PAGE (Mini-PROTEAN^® TGX™ #456109, Bio-Rad). Following migration, proteins were transferred onto 0.45 µm nitrocelullose membranes (GE Healthcare). Membranes were blocked in Tris-buffered saline (TBS) for 1 h at RT and then incubated overnight at 4°C with the primary antibody. After three washes in TBS with 0.05% Tween 20 (TBST) buffer for 10 min, the membrane was incubated with the appropriate HRP-conjugated secondary antibody (Jackson ImmunoResearch) diluted at 1:5000 in TBST-5% dry fat milk and signals were detected using chemiluminescence (see Table S4). Bands were then quantified by densitometry using ImageJ software and protein level expression was normalized with respect to the loading control protein.

Stage 31 living embryos were placed in anesthetic solution media [0.02% MS222 (Sigma, E105505) in 0.1× MBS] in a home-made 3D-printed Petri dish specially designed to maintain them with the dorsal side on the top. The setup was placed under compact Stereo Microscope ZEISS Stemi 305 coupled to ZEISS Axiocam 105 Color microscope camera. Timelapse recording was done after release of 1 µl of visible dyed microspheres (Bangs Laboratories, DSCR006; mean diameter 5.19 µm, diluted 1:2 in 0.1× MBS) at the anterior front of the embryo.

Cilia beating frequency was analyzed on the same pool of living embryos. Embryos were placed between a glass slide and a coverslip with a drop of anesthetic medium (0.02% MS222 in 0.1× MBS) surrounded by grease (high vacuum grease; DOW CORNING, Sigma-Aldrich Z273554-1EA) to stick the coverslip. Movies with a duration of 3 s (250 fps) were recorded at the ventral border of the embryos with a Nikon eclipse Ti-E microscope and a 63× long working distance air objective. Computation of cilia beat frequency (CBF) was done using an in-house routine developed in Python (Khelloufi et al., 2018). The resulting frequencies (Hz) shown in the quantification represent the mean CBF computed over all visible cilia per individual MCC. Movies are played at 50 fps to observe better the beating strokes and defaults. After flow and CBF recording, MO efficiency was assessed in those embryos by immunostaining for Centrin, acetylated α-Tubulin and mRFP (injection tracer).

Aeromonas hydrophila (Chester Stanier, ATCC^®7966 kindly provided by Eamon Dubaissi, University of Manchester, UK) colonies were grown for 48 h at 37°C on LB Agar plus 30 µg/ml kanamycin. Next, individual colonies were cultured in 10 ml LB plus 30 µg/ml kanamycin overnight at 37°C. The OD₆₀₀ (OD600 DiluPhotometer version1.4, Implen) was measured, and the culture was either diluted in LB or grown further to reach an OD₆₀₀ of 0.2. Cultures were centrifuged 10 min at 3500 g and the bacteria pellet was resuspended in the same volume of 0.1× MBS. Controls (non-injected or mRFP-injected) and morphant (Lrrcc1-MO plus mRFP, or Ccdc61-MO plus mRFP) embryos were incubated from stage 31 during 72 h at 13°C in 3 ml of bacteria-containing medium. Survival was assessed by recording active response to touch at 2 h, 4 h, 24 h, 48 h and 72 h.

At the end of the experiment, MO efficiency was assessed by immunostaining for Centrin, acetylated α-Tubulin and mRFP tracer.

For all experiments, statistical analysis of significance was performed using GraphPad Prism (version 8.2.0 for Windows, GraphPad Software, San Diego, USA). First, the normality of data (Gaussian distribution) was tested using a D'agostino and Pearson test (if n>50) or Shapiro–Wilk test (if n<50). When the data followed a normal distribution, we then compared them using parametric two-tailed unpaired Student's t-tests (between two groups) or one one-way ANOVA and Tukey HSD post-hoc test (between >2 groups). When data did not follow a normal distribution, we compared them using Mann–Whitney (2 groups) or Kruskal–Wallis (>2 groups) non-parametric tests. P=0.0001–0.001 or <0.0001 was considered extremely significant (***), P=0.001–0.01 was considered very significant (**); P=0.01–0.05 was considered significant (*) and P=≥0.05 was considered not significant (ns). Number of embryos and experiments performed for all analysis are listed in Table S5.

The authors wish to thank J. Azimzadeh, P. Walentek, E. Dubaissi and B. Mitchell for the gift of reagents. Imaging in IBDM was performed on PiCSL-FBI core facility, supported by the French National Research Agency through the program “Investments for the Future” (France-BioImaging, ANR-10-INBS-04). The authors thank Florian Roguet for Xenopus care, Rémi Flores-Flores for ImageJ macro development, Brice Detailleur for 3D printing of embryo holders, Fabrice Richard, Nicolas Brouilly and Aicha Aouane from IBDM electron microscopy facility.

The peer review history is available online at https://journals.biologists.com/jcs/article-lookup/doi/10.1242/jcs.258960.