ABSTRACT
The molecular mechanisms by which cilia orientation is coordinated within and between multi-ciliated cells (MCCs) are not fully understood. In the mouse oviduct, MCCs exhibit a characteristic basal body (BB) orientation and microtubule gradient along the tissue axis. The intracellular polarities were moderately maintained in cells lacking CELSR1 (cadherin EGF LAG seven-pass G-type receptor 1), a planar cell polarity (PCP) factor involved in tissue polarity regulation, although the intercellular coordination of the polarities was disrupted. However, CAMSAP3 (calmodulin-regulated spectrin-associated protein 3), a microtubule minus-end regulator, was found to be critical for determining the intracellular BB orientation. CAMSAP3 localized to the base of cilia in a polarized manner, and its mutation led to the disruption of intracellular coordination of BB orientation, as well as the assembly of microtubules interconnecting BBs, without affecting PCP factor localization. Thus, both CELSR1 and CAMSAP3 are responsible for BB orientation but in distinct ways; their cooperation should therefore be critical for generating functional multi-ciliated tissues.
INTRODUCTION
Multi-ciliated cells (MCCs), which have a few hundred cilia on their apical surface (Brooks and Wallingford, 2014; Meunier and Azimzadeh, 2016; Reiter and Leroux, 2017; Spassky and Meunier, 2017; Boutin and Kodjabachian, 2019), are specialized cells found in the epithelium of the oviduct, the trachea and the ependyma of the brain. In the infundibulum of the mouse oviduct (fallopian tube), eggs released from the ovaries are transported to the uterine side by the movement of multi-cilia (Shi et al., 2011; Li and Winuthayanon, 2016). In MCCs, the direction of cilia movement is coordinated both intracellularly (within each cell: rotational polarity) and intercellularly (between cells: tissue-level polarity) along the axis of the organ to generate directed flow (Hirota et al., 2010; Guirao et al., 2010; Mirzadeh et al., 2010; Tissir et al., 2010; Boutin et al., 2014; Meunier and Azimzadeh, 2016). Ciliary abnormalities, including motility and polarity as well as ciliogenesis in various organs, can lead to various disorders (Choksi et al., 2014; Reiter and Leroux, 2017).
This unidirectional ciliary movement is reinforced by the structural polarity of each cilium. The basal body (BB) found at the base of the cilium has similar components to the centriole and supports the axoneme in the apical cytoplasm. The basal foot (BF) is an electron-dense structure that projects unidirectionally from the BB (Werner et al., 2011; Kunimoto et al., 2012; Reiter et al., 2012; Clare et al., 2014; Garcia and Reiter, 2016) in the same orientation as the effective stroke of the cilium.
The orientation of multiple cilia reflects the polarity of each cell. In epithelial cells, in addition to apicobasal polarity, planar cell polarity (PCP) runs perpendicular to the apicobasal plane within the plane of the epithelium (Aw and Devenport, 2017; Butler and Wallingford, 2017; Henderson et al., 2018). PCP factors, which include CELSR1 (cadherin EGF LAG seven-pass G-type receptor 1) and the Van Gogh-like proteins VANGL1 and VANGL2 are proteins involved in regulating PCP that were initially identified by genetic screens performed in Drosophila melanogaster; many are conserved in vertebrates regulating morphological polarities extending over multiple cells (Boutin et al., 2012; Shi et al., 2013; Devenport, 2014; Hale and Strutt, 2015; Yang and Mlodzik, 2015; Koyama et al., 2016; Shi et al., 2016a; Butler and Wallingford, 2017; Koyama et al., 2019). These factors localize asymmetrically at cell boundaries in each cell according to the tissue's polarity (Park et al., 2008; Mitchell et al., 2009; Hirota et al., 2010; Ohata et al., 2014). A cytoplasmic PCP factor, DVL2 (dishevelled2), also localizes to the base of cilia, in addition to cell boundaries in Xenopus laevis epidermis and mouse ependyma. PCP factors are essential for regulating cilia polarity in Xenopus epidermis (Park et al., 2008; Mitchell et al., 2009), mouse ventricular ependymal cells (Guirao et al., 2010; Hirota et al., 2010; Mirzadeh et al., 2010; Tissir et al., 2010; Boutin et al., 2014; Ohata et al., 2014; Takagishi et al., 2017, 2020), mouse trachea (Vladar et al., 2012) and mouse oviduct (Shi et al., 2014). We previously reported that Celsr1-deficient females are sterile because of multiple PCP defects in the mutant oviducts (Shi et al., 2014). By using fluid flow analysis and direct observation of ciliary movement, we found that the directions in which the cilia of MCCs move were not aligned along the longitudinal axis of the oviduct. Our analysis following electron microscopy observation has revealed that cilia orientation is not coordinated in Celsr1−/− mutant oviducts. However, we could only observe details in a very limited area, and could not simultaneously determine BB orientation across multiple cells.
Cytoskeletal elements, namely microtubules, intermediate filaments and actin filaments, are connected to the BB (Sandoz et al., 1988; Werner et al., 2011; Clare et al., 2014; Antoniades et al., 2014; Chevalier et al., 2015; Herawati et al., 2016; Tateishi et al., 2017) and supposedly provide mechanical support for cilia. The mechanism by which these cytoskeletal elements are involved in cilia polarity regulation differs between animal species and between organs (Iftode and Fleury-Aubusson, 2003; Werner et al., 2011; Antoniades et al., 2014; Bayless et al., 2019; Soh et al., 2019). In mouse trachea, actin filaments function in BB docking to the apical surface of cells, and microtubules regulate cilia orientation (Herawati et al., 2016). Electron microscopy has revealed that microtubules accumulate near the tip of the BF in various animal species (Reed et al., 1984; Sandoz et al., 1988; Lemullois and Marty, 1990). In addition, microtubules are concentrated at the apical cell–cell boundary in a polarized manner in MCCs of mouse ependyma and trachea; this polarized enrichment correlates with the orientation of BBs (Vladar et al., 2012; Boutin et al., 2014; Herawati et al., 2016; Takagishi et al., 2017). We have also previously shown that EB1, which localizes to the plus-end of microtubules, is enriched in a polarized manner in mouse oviduct MCCs (Shi et al., 2016b). The molecular mechanisms underlying microtubule control of cilia orientation remain unclear.
In this study, we explored how the orientation of cilia is coordinated within each cell and between cells, focusing on MCCs in the infundibulum of the mouse oviduct. To this end, we made use of super-resolution microscopy, stimulated emission depletion microscopy (STED), to dissect the roles of PCP factors and microtubules in establishing tissue-wide coordination of cilia orientation. In addition to the role of CELSR1, a PCP factor that organizes the orientation of cilia and polarity of microtubule enrichment at the tissue level, we found that CAMSAP3 [calmodulin-regulated spectrin-associated protein 3; also known as nezha (Meng et al., 2008)], a microtubule minus end-binding protein, localizes to BBs in a polarized manner, controls rotational polarity of cilia and organizes the assembly of microtubules interconnecting BBs.
RESULTS
Intracellular basal body orientation is moderately coordinated in Celsr1-deficient oviduct MCCs
In the Celsr1−/− mutant, both ODF2 and centriolin signals were detected in each BB with the same shape and in the same relative positions as in WT. The average number of BBs in adult Celsr1−/− mutant cells was 199 per cell (n=20 cells, s.d.=27.7), which was comparable to WT. The number of BBs for which we could analyse the orientation was less in the Celsr1−/− mutant cells (127 BBs per cell on average; n=20 cells; Fig. 1B′,C′) than in WT, because the apical surface was not flat but dome shaped, and it was not easy to capture the orientation of cilia over a wide area in individual cells. The variation in BB orientation within each Celsr1−/− mutant cell was compared with WT. The CV values for the Celsr1−/− mutants were between 0.09 and 0.5 (median=0.23), which was significantly higher than WT, but still lower than completely random (Fig. 1D; Fig. S5). In the Celsr1−/− mutant, the mean BB orientation varied between cells. This was the case even when BB orientation was approximately aligned with lower CV values in neighboring cells (Fig. 1E′, purple and blue arrows).
These analyses confirmed that BBs in each cell are not randomly oriented in WT, and Celsr1−/− mutants show reduced coordinated orientation of BBs, and the mean BB orientation differed between cells. These results are consistent with our previous finding (Shi et al., 2014) and suggest that CELSR1 is involved in coordinating cilia orientation between cells, although other unidentified mechanisms appear to exist for aligning cilia orientation in each cell, even when CELSR1 is absent.
Polarized microtubule enrichment at the cell–cell boundary correlates with BB orientation in each cell even in the absence of CELSR1
We next analysed the relationship between the orientation of microtubule enrichment and BB orientation in Celsr1−/−. BB orientation was determined by ODF2 and centriolin signals as shown in Fig. 1. As the optimal fixation conditions for the anti-β-tubulin antibody and the anti-ODF2 and anti-centriolin antibodies are different, the respective staining signals were slightly different from that which is described above (Fig. 1A; Fig. 2A′). The β-tubulin signal was observed in the form of fine dots, nevertheless allowing us to observe its density gradient (Fig. 2C). The BB orientation was determined and quantified in cells where microtubule enrichment was observed (Fig. 2C′). These cells showed CV values lower than 0.25 (median=0.14), and the mean BB orientation (purple arrow in Fig. 2C′) in each cell coincided with the microtubule enrichment direction (yellow arrow in Fig. 2C; n=5 cells, average 210 BBs per cell).
Thus, microtubules were enriched near the apical surface of MCCs even in the absence of CELSR1, and the orientation of microtubule enrichment correlated with the mean BB orientation in each cell; however, microtubule enrichment orientation between cells was not aligned. These results indicate that BBs are oriented towards the region of microtubule enrichment near the apical surface, and CELSR1 plays a role in aligning the orientation of microtubule enrichment between cells.
Connection between microtubules and BBs in oviduct MCCs
Electron microscopic observation of MCCs has previously shown that microtubules are attached to the tip of the BF and run along the apical surface interconnecting adjacent BF (Reed et al., 1984; Sandoz et al., 1988; Vladar et al., 2012; Herawati et al., 2016; Tateishi et al., 2017). It remains unclear, however, how the microtubules are connected to the BBs and how they are involved in organizing BB orientation. In order to determine microtubule localization and elucidate the relationship between microtubules and BB and BF in detail, microtubules and BBs were visualized using an anti-β-tubulin antibody and GFP-centrin2, respectively. Microtubules formed a striped pattern on the plane parallel to the apical surface in the range of 0.6 µm, which was identical to GFP-centrin2 (Fig. 2D; Movie 1). The signal intensities of the stripes were also stronger on the uterine side of each cell (Fig. 2D) and weaker on the ovarian side. When examined from the lateral side of the cells, a thick microtubule bundle (visualized with anti-α-tubulin antibody) connected to each BB (visualized with GFP-centrin2) extended to the basal side of the cell in the cytoplasm (Fig. 2E), although which components of microtubules visualized by lateral sections correspond to the apical microtubule stripes remains unclear. We also examined Celsr1−/− oviducts and found that the apical microtubule stripes were maintained, while apico-basal bundles were clearly connected to BBs (Fig. S1B,C). To summarize, oviduct MCCs have microtubules running along BBs as well as those extending basally from BBs, and either one of them, or both, are enriched at a cell boundary; CELSR1 is not involved in the assembly of these microtubules, although it is important for their PCP orientation.
CAMSAP3, a microtubule minus-end binding protein, localizes asymmetrically to the base of cilia
Although γ-tubulin localizes in BBs and BF (Muresan et al., 1993; Hagiwara et al., 2000), it is unknown if γ-tubulin regulates ciliary orientation (Yuba-Kubo et al., 2005). The minus-ends of axonemal microtubules are oriented towards the BBs in MCCs. In intestinal non-ciliated epithelial cells, the minus-ends of non-centrosomal microtubules that run apico-basally are anchored to the apical surface by CAMSAP3 (Meng et al., 2008), a non-centrosomal microtubule minus-end binding protein (Toya et al., 2016). Based on the above, we sought to determine if CAMSAP3 plays a role in MCCs and if it is involved in organizing microtubules to regulate the orientation of BBs.
The relative positions of CAMSAP3 and other BB and BF markers in mature MCCs are summarized in Fig. 3E. We observed that CAMSAP3 was located in the same direction as the BF from the apical view, but was located apically when compared with other BF markers such as centriolin.
CAMSAP3 protein localizes to the BB prior to coordination of BB orientation
Next, we analysed when CAMSAP3 became co-localized with the BB components during the development of MCCs. In P13 oviducts, CAMSAP3 showed dot-like signals around the apical surface in all cell types (Fig. 4B). This apical localization of CAMSAP3 in the oviduct MCCs was similar to that reported in adult nasal respiratory epithelia, but we did not observe CAMSAP3 localized in two layers during the early stage of development as reported (Robinson et al., 2020). In type IV and type V cells, CAMSAP3 was observed adjacent to GFP-centrin2. CAMSAP3 signal intensity was stronger in type IV than in type V cells, and CAMSAP3 dots were widely distributed over the apical surface while GFP-centrin2 formed clusters of dots. In type V cells, CAMSAP3 localized at BBs, which was similar to their observed localization in adult MCCs, while CAMSAP3 dots were even smaller when compared with those in adult MCCs. These results suggest that CAMSAP3 localizes to BBs both during and after BB docking to the cell surface.
We also examined the localization of ODF2 and centriolin using P13 mouse oviducts (Fig. S3E,F). Centriolin colocalized with GFP-centrin2 in all cell types; however, ODF2 and centriolin colocalized only in type IV and type V cells (Fig. S3F). ODF2 localized to BBs in 83% of type IV cells, and 100% of type V cells. The above results indicate that CAMSAP3 localizes to the BB at a later time than other BB and BF components, including centrin2, centriolin and γ-tubulin, which were already colocalized before BB docking. ODF2, however, localizes to BBs later than CAMSAP3.
We next focused on type IV and type V cells and determined BB orientation (Fig. 4C–E). In type IV cells, individual BBs were oriented in various directions within each cell, and CV values varied widely between 0.2 and 0.9 (n=28 cells, median=0.53). In type V cells, however, BBs were oriented in a similar direction, and the CV value of most cells was smaller than 0.32 (n=28 cells, median=0.13) (Fig. 4E). Thus, BB orientation was aligned during the transition from type IV to type V. Taken together, these observations revealed that CAMSAP3 localizes to the BBs during their docking to the plasma membranes prior to the coordination of BB orientation.
BB orientation is not coordinated in Camsap3dc/dc mutants
Electron microscopy observations also revealed BB orientation defects in the Camsap3dc/dc mutant (Fig. 5E,E′), and were consistent with the results obtained using super-resolution microscopy. Similar results were recently reported in mouse nasal epithelium and in Xenopus after Camsap3 gene inactivation (Robinson et al., 2020). In a minor population, we observed structures similar to the BF protruding from multiple positions on BBs (Fig. S4A). Pronounced defects in assembly of the central pair were reported in the nasal MCCs (Robinson et al., 2020), although such abnormalities were much less evident in the oviduct MCCs (Fig. S4B). These results indicate that CAMSAP3 is indispensable for aligning BB orientation in each cell.
Planar polarized protein localization is maintained in Camsap3dc/dc
Apical microtubule distribution is disturbed in Camsap3dc/dc mutant MCCs
To elucidate the mechanism of how CAMSAP3 coordinates cilia orientation within cells, we examined the distribution of microtubules more closely in Camsap3dc/dc mutant MCCs, as it might be related to the disorganization of BB orientation. In Camsap3dc/dc mutants, α-tubulin was distributed to fill the spaces between GFP-centrin2 signals as observed in WT. Notably, however, microtubule stripes became fragmented in the mutant cells (Fig. 6E′), although some of BBs were still aligned in rows (Fig. 6E,E′, dashed lines). The analysis of the overall staining intensity of α-tubulin located at the level of GFP-centrin2 in individual cells (in the range of 0.6 µm) showed that, while α-tubulin was enriched at the uterine side of WT cells, this enrichment was no longer observed in Camsap3dc/dc cells (compare α-tubulin panels in Fig. 6F,F′), although other microtubule populations still exhibited a gradient distribution as shown in Fig. 6C. We next examined the microtubules, which extend apico-basally into the cytoplasm from BBs, from the lateral side (Fig. 6G,G′), finding that the assembly of this population of microtubules looked normal in Camsap3dc/dc mice. We also checked for acetylation of microtubules by staining with anti-acetylated tubulin antibody, but we did not detect any differences between WT and mutant mice (Fig. S4E,E′). These observations suggest that the assembly of microtubules horizontally associated with BBs are selectively regulated by CAMSAP3.
As a previous study showed that CAMSAP3 binds actin filaments via ACF7 (also known as MACF1; Ning et al., 2016), we observed actin distribution using phalloidin. Phalloidin signals also formed stripes about 0.2 µm basal from the plane of α-tubulin stripes in WT (Fig. 6E), whereas in Camsap3dc/dc, the distributions of α-tubulin and phalloidin signals became less identical (Fig. 6E′, arrowheads). At the apical surface of MCCs, phalloidin-positive protrusions were observed more frequently in the mutant cells (Fig. S4C,C′). A small number of these protrusions showed a cyst-like structure, and the cysts occasionally contained microtubules or multiple axonemes (Fig. S4C′,D), which are never seen in WT cells. These suggest that actin filament organization was also affected as a result of Camsap3 mutation. Finally, we observed the distribution of CAMSAP3 in Celsr1−/− MCCs, and found no change in its distribution after CELSR1 loss (Fig. S6A,A′), implying that the CELSR1 and CAMSAP3 systems differentially regulate BB orientation.
DISCUSSION
The present study analysed BB orientation and its coordination in the oviducts using two mutants which have lost CELSR1 or CAMSAP3 functions. Each mutant displayed distinct abnormalities concerning these processes. In Celsr1 mutants, PCP was partly abrogated, whereas the intracellular polarity represented by BB orientation was moderately maintained within individual cells. Camsap3-mutated cells showed disruption of BB orientation in individual cells, but not losing PCP of the cell sheet. These observations suggest that BB orientation in individual cells occurs in the absence of CELSR1, and cooperation of the CELSR1 and CAMSAP3 systems is necessary to determine the entire polarity of oviducts at both cellular and tissue levels.
CELSR1 coordinates cilia orientation between cells in the oviduct
Fluid flow in the oviduct/fallopian tube was recently reported to be directed from the uterine side to the ovarian side by the secretion of fluid and contractility of smooth muscles in the isthmus of the oviduct (Hino and Yanagimachi, 2019). Eggs are transported steadily against the direction of fluid flow from the fimbria to the ampulla by directional movements of the cilia, which are coordinated both within individual cells and between cells. In this study, observing BBs with super-resolution microscopy (STED) allowed us to determine whether cilia orientation is coordinated between cells even in the wide field within the folds of the infundibulum epithelium. We confirmed that cilia orientation is aligned with organ orientation, i.e. the longitudinal axis of the oviduct in the infundibulum.
The mechanism for coordinating the orientation of multi-cilia differs between animal species and between organs. Boutin and colleagues have reported that the functions of PCP factors also vary within a tissue (Boutin et al., 2014). In the mammalian ependyma, one group of PCP factors [CELSR2, CELSR3, FZD3 (frizzled-3) and VANGL2] regulates cilia orientation within individual cells, while another group (CELSR1, FZD3 and VANGL2) regulates cilia orientation between cells.
The involvement of CELSR1 in the coordination of cilia orientation may differ between ependyma and oviduct epithelium. While cilia patch displacement and microtubule enrichment were correlated in Celsr1−/− ependyma, the orientation of patch displacement did not correlate with mean cilia orientation. In contrast, the direction of microtubule enrichment was not coordinated between oviduct cells in Celsr1−/−, while the mean BB orientation coincided with the direction of microtubule enrichment. Among the CELSR family proteins, CELSR1 probably plays a major role in coordinating cilia orientation in the oviduct epithelium because CELSR3 expression was not detected, and Celsr2 mutants have no obvious ciliary phenotype in the oviduct. Cilia orientation was moderately coordinated in individual cells lacking CELSR1, but the orientation between cells was not the same. This suggests that CELSR1 is mainly responsible for coordinating cilia orientation between cells.
We cannot rule out the possibility that other PCP factors may coordinate cilia orientation within each cell. However, even when Vangl2 is deficient, oviduct MCCs do not show visible abnormalities (F.M.U., M.A. and T.F., unpublished data), in contrast to ependymal MCCs. We also have previously reported that in Celsr1−/− mutants, VANGL2 proteins weakly localized on the cell membrane, but the distribution was more uniform (Shi et al., 2014), suggesting that the function of VANGL2 is also partly affected. FZD6 was distributed unevenly in Celsr1−/− oviduct MCCs (data not shown), and this localized FZD6 may be regulating intracellular BB organization. Another possibility is that intracellular coordination of cilia orientation in oviduct MCCs is less dependent on PCP factors.
Microtubule networks and the regulation of cilia orientation in individual cells by CAMSAP3
Cells that lack CELSR1 were proposed to retain the ability to self-organize the distribution and orientation of cilia within individual cells. In mammalian MCCs found in the ependyma and trachea, microtubules are considered important for aligning cilia orientation within the cells (Clare et al., 2014; Herawati et al., 2016; Tateishi et al., 2017). These microtubules that localize just below the apical surface are connected to the BF. Microtubules are also enriched on one side of cells in a region that is slightly deeper than the apical microtubule meshwork around the cell–cell boundaries. We found that this is also common in oviduct MCCs. Moreover, we observed that microtubules are concentrated on the uterine side of each cell, and the apical microtubule meshwork forms a striped pattern with a gradient from the uterine side along the longitudinal axis of the oviduct. We found that in the Celsr1−/− mutant, microtubule stripes were present between the BBs (Fig. S6B), and microtubule enrichment was also conserved in more than 75% of MCCs. BBs were oriented towards the site of microtubule enrichment in these cells. These results suggest that CELSR1 is not essential for intracellular microtubule enrichment and apical stripe formation.
We found that CAMSAP3, a non-centrosomal microtubule minus-end binding protein, localizes to the base of the cilia of MCCs as recently reported (Robinson et al., 2020), in addition to γ-tubulin, a centrosomal microtubule minus-end binding protein (Clare et al., 2014). γ-tubulin has been shown to be localized in the BF, and we found that CAMSAP3 also exists asymmetrically in the same direction as γ-tubulin in each BB in the plane that is more apical to the position of γ-tubulin. The apical microtubule stripes and CAMSAP3 are on the same plane, and CAMSAP3 may interact with microtubules in this plane. During the maturation of MCCs, γ-tubulin exists in the BB regardless of the stage, but CAMSAP3 only localizes to BBs at the stage of BB docking onwards. Whereas the subcellular localization of γ-tubulin, centrin2 and centriolin were similar at all stages even when BBs were within the cytoplasm, CAMSAP3 was present on the apical region throughout all stages. From the above, it is suggested that γ-tubulin and CAMSAP3 may play different roles in organizing microtubules in MCCs. In intestinal epithelial cells that do not have multiple cilia, the absence of CAMSAP3 affected microtubules which run from the basal and attach to the apical surface (Toya and Takeichi, 2016), whereas the microtubules running apicobasally were maintained in Camsap3dc/dc mutant oviduct MCCs. However, the distribution of apical microtubules around BBs was disturbed. The intracellular orientation of cilia was not aligned in the oviduct, as reported in nasal respiratory epithelia (Robinson et al., 2020). This suggests a functional link between cilia orientation within individual cells and microtubule distribution in the apical surface. We speculate that the minus-end of microtubules running apicobasally may be bound to γ-tubulin in MCCs, but this needs to be elucidated in the future.
Why is cilia orientation disturbed in the Camsap3 mutant?
Intracellular BB orientation was disorganized within individual cells in the Camsap3 mutant. It is unlikely that the cells lacking CAMSAP3 disrupt planar polarity. Aside from cilia orientation, other patterning elements look undisturbed including localization of PCP factors, enrichment of microtubules on the cell edge and folds along the longitudinal axis of the oviduct.
Both CAMSAP3 and ODF2 may become localized to BBs when the polarity or asymmetry of BBs is established. In Odf2 mutant tracheal MCCs, the BF structure is also lost, and ODF2 is essential for the polar formation of BBs (Kunimoto et al., 2012). CAMSAP3 is, however, dispensable for the BF formation, because the structure of BBs and the BF is almost normal except for some minor populations in Camsap3dc/dc. BB/BF components (ODF2, centriolin, centrin2, γ-tubulin) localize to BBs in the absence of CAMSAP3. We observed some BBs with multiple BF only in Camsap3 mutants; these BBs look similar to ‘hybrid type BBs’ reported recently (Liu et al., 2020), and may be increased in the Camsap3-deficient conditions. Otherwise, CAMSAP3 may also be responsible for limiting BBs to a single BF.
The BB orientation defect is probably due to local abnormalities around BBs. To align BBs and coordinate orientation of BBs, BF should be connected in the same orientation by cytoskeletal elements. One possibility is that CAMSAP3 is involved in anchoring BF to microtubule networks. We observed the impaired accumulation of microtubules in the apical region around BBs in Camsap3dc/dc mutant MCCs. We speculate that stable apical microtubule stripe formation is disrupted or the stability of apical microtubules are defective in the absence of CAMSAP3. Due to this change, the BF cannot anchor stably to the cytoskeleton, and the orientation of BBs are not aligned. The binding of CAMSAP3 and ACF7 may mediate the interaction between microtubules and actin filaments (Ning et al., 2016; Noordstra et al., 2016) and this interaction may also help microtubules to accumulate around BBs and connect neighboring BF in the same orientation. The stable accumulation of apical microtubules connecting the BF may therefore be the major function of CAMSAP3.
The mechanism for aligning cilia orientation within and between cells
In this study, we propose that the orientation of multi-cilia is regulated intracellularly and intercellularly. The direction of cilia has been suggested to be autonomously aligned within individual cells, regulated by CAMSAP3. CELSR1 is a PCP factor that aligns microtubule networks between cells to coordinate tissue-wide cilia orientation. Thus, both CELSR1 and CAMSAP3 systems are necessary to organize cilia orientation within a tissue. CAMSAP3 may be responsible for integrating polarity signals locally around the BBs downstream from PCP signals including CELSR1 via microtubules, as discussed in a previous report (Kunimoto et al., 2012). We tried to analyse interaction between these two mechanisms by making double knockouts, but we could not obtain the Camsap3dc/dc–Celsr1−/− mice because they were perinatal lethal. In addition to these two hierarchies, microtubules are enriched at cell boundaries in a polarized fashion and form a concentration gradient inside cells, and these may connect the two systems. How these multi-scale mechanisms are connected so that the orientation of cilia in the whole tissue is aligned should be elucidated in the future.
MATERIALS AND METHODS
Animals
Female mice (Slc:ICR; SLC, Japan) were used to study MCCs in WT. Celsr1−/− mutant mice (Ravni et al., 2009), Camsap3dc/dc mutant mice (Toya et al., 2016) and GFP-Centrin2 (GFP-CETN2) Tg mice (a gift from Holden Higginbotham, Brigham Young University - Idaho, Rexburg, ID, USA; transferred from Hiroshi Hamada, RIKEN Center for Biosystems Dynamics Research, Japan; Higginbotham et al., 2004) were described previously. Animal care and experiments were conducted in accordance with the Guidelines of Animal Experimentation of National Institutes for Natural Sciences and RIKEN animal experimentation guidelines. All animal experiments were approved by either the Animal Research Committee of National Institutes for Natural Sciences or by the Institutional Animal Care and Use Committee of Riken Kobe Branch, respectively. Animals were maintained in a light- and temperature-controlled room using a 12 h:12 h light:dark cycle at 23±2°C.
Immunohistochemistry
The fluorescent images were taken using a Nikon A1 confocal microscope and TCS SP8 STED (Leica). Sample preparations were basically the same as previously reported (Shi et al., 2014). Anti-ODF2 antibody-producing hybridoma was a gift from Sachiko Tsukita, Osaka University, Japan (Tateishi et al., 2013). All primary antibodies and secondary antibodies used in this study are listed with fixation conditions in Table S1.
After 4% PFA fixation, samples were treated with 0.1% Triton X-100 in PBS at room temperature (RT) for 1 h. Fixed samples were blocked with Blocking One (Nacalai Tesque, Kyoto, Japan) at RT for 1 h, and incubated with primary antibodies in Blocking One at 4°C overnight. Incubation with α- or β-tubulin antibodies was for 36 h at 4°C. After washes with PBS, samples were incubated with secondary antibodies at RT for 3 h or 4°C overnight, or 24 h at 4°C for α- or β-tubulin antibody staining. After washes, each sample was mounted in Fluoromount-G (SouthernBiotech, Birmingham, AL, USA). ProLong Diamond (Thermo Fisher Scientific) was used to mount samples for STED observation. Detailed conditions for STED observation are shown in Table S1. The ovarian side of the oviduct was determined by the presence of fimbriae.
Quantifying the localization of PCP proteins and microtubule enrichment
Localization of PCP proteins was quantified by calculating nematic order as described (Aigouy et al., 2010; Shi et al., 2014). Maximum intensity projection (MIP) images were rendered from five serial images acquired with a step size of 1 µm. The orientation of microtubule enrichment was determined by using this method with minor modifications. As microtubules were enriched at the proximity of cell boundaries, cell boundaries were labeled with phalloidin and microtubule signals within the range of 0–5 pixels away from the cell boundaries were extracted using ImageJ. Nematic order defines only the axis and the magnitude of polarity. Additionally, the vector of microtubule polarity was determined by separating microtubule signals into two halves with a line perpendicular to the axis of the polarity, and by comparing intensities between these two separated microtubule signals. MIP images were rendered from 20 serial images acquired with a step size of 0.2 µm.
Quantifying basal body orientation
BB orientation was determined by ODF2 and centriolin signals after acquiring images with TCS SP8 STED (Leica). The detailed conditions for STED imaging are provided in Table S2. All images were acquired at a step size of 0.1 µm, and images from the ODF2 channel were subjected to an ImageJ macro (Correct 3D Drift). The same parameter of ‘Correct 3D drift’ was applied for images of centriolin. On each ODF2 ring, an arrow pointing to the center of ellipse/two points of centriolin signals (as shown in Fig. 1A′) was manually plotted. Making use of shape anisotropy of the arrow, the angle between the arrow and the longitudinal axis (ovary>uterus) of the oviduct was measured by Shape Descriptor Map (BioVoxxel_Toolbox; Fiji). In each cell, the CV of each set of angles was calculated using R as described previously (Zar, 2010; Shi et al., 2014). To acquire images using a transmission electron microscope (JEM-1010, Jeol; equipped with 2k×2k CCD camera, Olympus Soft Imaging Solutions), sample blocks were prepared as described (Shi et al., 2014). Ultrathin sections (50–70 nm) were made with a microtome (Leica EM UC7), and images were obtained on a grid (OkenShoji150-D, 75-A mesh). To measure the orientation of BB in transmission electron microscope images, an arrow was drawn manually from the center of BB to BF. CV was calculated as described above.
Time-lapse recordings of MCC maturation
The oviducts of GFP-CETN2 Tg mice (postnatal day 13) were opened longitudinally, and placed in a drop of Dulbecco's modified Eagle's medium with high glucose (Gibco#31053) containing 10% FBS on a glass bottom dish (MatTek P35G-1.5-10-C). The drop was covered with liquid paraffin (128-04375, Wako) to avoid evaporation. The apical surface of the luminal epithelium faced the bottom of the dish. Oviducts were cultured at 37°C in 5% CO2 during the recording, and imaged with a spinning disc confocal system (Cell Voyager CV1000; Yokogawa) using a 100× objective lens (UPLSPAPO100XS, Olympus) at 1 h intervals. At each time point, 40 slices were acquired at a step size of 0.5 µm.
Acknowledgements
We thank Holden Higginbotham and Hiroshi Hamada for the GFP-Centrin2 mouse; Sachiko Tsukita for the anti-ODF2 antibody and hybridoma; Masahiko Itoh and Mikio Furuse for the anti-ZO1 antibody; Yoshikatsu Sato and Tetsuya Higashiyama (ABiS) for technical support and advice concerning STED microscopy; Gen Yamada for transmission electron microscope support; and Hiroko Saito, Azusa Kato, Tetsuhisa Otani and other laboratory members for technical support with animal care and helpful discussions. We also thank ExCELLS, the Spectrography and Bioimaging Facility (NIBB) and the EM facility (NIPS) for allowing us to use STED, confocal microscopes and an electron microscope, respectively.
Footnotes
Author contributions
Conceptualization: F.M.U., M.A., T.F.; Methodology: F.M.U., M.A., D.S., T.F.; Software: M.A., D.S.; Investigation: F.M.U., D.S., Y.H.; Resources: S.O., Y.H., F.T., M.T.; Writing - original draft: F.M.U., M.A., M.T., T.F.; Writing - review & editing: F.M.U., M.A., T.F.; Visualization: F.M.U., M.A.; Supervision: M.T., T.F.; Project administration: T.F.; Funding acquisition: M.A., T.F.
Funding
This work was supported by the Japan Society for the Promotion of Science, KAKENHI (grant numbers 25291054, 15H01220, 17H03689, 16H06280 and 19K16153) and the Japan Science and Technology Agency, Core Research for Evolutional Science and Technology (JPMJCR1654 to T.F.).
References
Competing interests
The authors declare no competing or financial interests.