Calaxin is required for cilia-driven determination of vertebrate laterality

Calaxin is a Ca2+-binding dynein-associated protein that regulates flagellar and ciliary movement. In ascidians, calaxin plays essential roles in chemotaxis of sperm. However, nothing has been known for the function of calaxin in vertebrates. Here we show that the mice with a null mutation in Efcab1, which encodes calaxin, display typical phenotypes of primary ciliary dyskinesia, including hydrocephalus, situs inversus, and abnormal motility of trachea cilia and sperm flagella. Strikingly, both males and females are viable and fertile, indicating that calaxin is not essential for fertilization in mice. The 9 + 2 axonemal structures of epithelial multicilia and sperm flagella are normal, but the formation of 9 + 0 nodal cilia is significantly disrupted. Knockout of calaxin in zebrafish also causes situs inversus due to the irregular ciliary beating of Kupffer’s vesicle cilia, although the 9 + 2 axonemal structure appears to remain normal.

M otile cilia and flagella are organelles that have been conserved through evolution [1][2][3] . They possess internal cytoskeletal structures, axonemes, that are composed of nine outer doublet microtubules and two central singlet microtubules (9 + 2 structure) 4,5 . Two types of projection extend from each microtubule doublet, the outer and inner dynein arm (ODA and IDA), both of which are large, multi-subunit complexes consisting of heavy, intermediate, and light chains. Dynein heavy chains (HCs) are motor subunits that hydrolyze ATP to convert chemical energy into mechanical energy for microtubule movement. The intermediate and light chains (ICs and LCs) assemble and regulate the motor subunits.
Genetic defects of the dynein components cause primary ciliary dyskinesia (PCD), a human ciliopathy disease [6][7][8] . PCD is characterized by defects in the motility of cilia and flagella in a variety of cells, including sperm, and in tissues of the trachea, ependyma, and embryonic node. By utilizing both mice and zebrafish as model systems, it is possible to acquire important insights into the phenotypes and mechanism of PCD [9][10][11] .
The motility of cilia and flagella is modulated in response to several extracellular stimuli [26][27][28] . The most critical intracellular factor mediating these changes is Ca 2+ . Calaxin is a neuronal calcium sensor protein first described in the sperm of the ascidian Ciona intestinalis [29][30][31] . It directly binds to the β-type heavy chain (orthologous to Chlamydomonas γ heavy chain 3 ) of the ODA in a Ca 2+ -dependent manner and regulates the propagation of the asymmetric flagellar wave. It is also necessary for changes in swimming direction during sperm chemotaxis 30 (Fig. 1a). In sea urchin embryos, calaxin is a critical regulator for the coordinated movements of monocilia and is a prerequisite for the establishment of ciliary orientation 32 (Fig. 1a), which is generally thought to be determined by the planar cell polarity. Calaxin is a Ca 2+ sensor that has evolved in the opisthokont (animal + fungi) lineage 3 ; however, it has not been widely studied, particularly in vertebrates.
The initial aim of this study was to elucidate the role of calaxin in vertebrate male fertility, particularly how sperm chemotaxis contributes to the success of vertebrate internal fertilization. We generated knockout mice lacking the gene encoding calaxin, Efcab1. Both male and female Efcab1 −/− mice were unexpectedly fertile. However, many Efcab1 −/− mice showed hydrocephalus and visceral inversion, both of which are typical features of ciliopathy, without apparent changes in 9 + 2 axonemal structures. Intriguingly, calaxin knockout caused a drastic loss of nodal cilia, whereas other cilia and flagella were normally formed.

Results
Calaxin knockout mice exhibit PCD but are fertile. To elucidate the physiological function of calaxin in vertebrates, we generated a knockout mouse in which exon 4 of Efcab1, the gene encoding calaxin, was genetically disrupted by homologous recombination (Supplementary Fig. 1a-c). No calaxin expression was observed in the sperm, trachea or ependyma of Efcab1 −/− mice either at mRNA ( Supplementary Fig. 1c) or at protein level (Fig. 1b, c,  Supplementary Fig. 1d). Notable phenotypes of postnatal Efcab1 −/− mice were hydrocephalus and situs inversus in 35% and 49% of offspring, respectively (Fig. 1d, e), with both phenotypes present in 16% of offspring (Fig. 1f).
In Ciona, calaxin is essential for chemotaxis of the sperm to the egg; however, both male and female Efcab1 −/− mice were fertile, although litter sizes when either or both parents were Efcab1 −/− were significantly lower compared with litters from wild-type parents (Fig. 1g). The litter size from Efcab1 −/− males and females was 1~2, whereas that from wild-type mice was 7~8. However, Efcab1 −/− sperm showed the same in vitro fertilization rate as wild-type sperm (Fig. 1h). The number of Efcab1 −/− embryos was almost the same as that of wild-type embryos at embryonic day 8 (E8) but after E14 the number of Efcab1 −/− embryos declined (Fig. 1i). Efcab1 −/− male and female offspring were born at non-Mendelian frequency with a deficit of Efcab1 −/− mice (Supplementary Table 1), indicating homozygotic embryonic lethality. Efcab1 −/− mice showed both cardiac hypertrophy (Fig. 1j) and enlargement of brain ventricles (Fig. 1d, k). Hydrocephalus emerges after birth and is not always lethal 33 ; therefore, the cause of embryonic lethality is most likely to result from cardiac defects that often accompany PCD 34 . Surviving Efcab1 −/− mice showed a similar survival rate to wildtype mice (more than 10 weeks) (Fig. 1l). A bacterial artificial chromosome containing the entire Efcab1 gene ( Supplementary  Fig. 2a, b) rescued the expression of calaxin protein (Supplementary Fig. 2c) and the PCD phenotypes of Efcab1 −/− mice ( Supplementary Fig. 2d), clearly indicating the distinct roles of calaxin in ciliary function.
Tracheal cilia of Efcab1 −/− mice also exhibited active motility ( Supplementary Movies 3 and 4). However, analysis of fluid flow in trachea using fluorescent beads revealed that the flow velocity in Efcab1 −/− mice was decreased to almost half that of wild-type mice (Fig. 3c,

Wild-type
Efcab1 -/-  embryonic node is critical for left-right asymmetry determination. This flow is generated by the rotary movement of monocilia. Bending of each cilium was clearly observed during the rotation (Supplementary Movie 13). To investigate the cause of situs inversus in Efcab1 −/− mice, we observed the nodal cilia. Compared with wild type embryos at E7.5, the nodal cilia of Efcab1 −/− embryos were strikingly sparse or completely absent in some cases and longer microvilli were more prominent on the cell surface (Fig. 4a).
The number of motile cilia on the node was less than 15% of the wild-type number ( Fluorescent microbeads at the Efcab1 −/− node showed a certain but random flow (Fig. 4d) and the velocity was significantly lower than that at the wild-type node (Fig. 4e). Next, we recorded and analyzed the trajectories of nodal cilia tips. The nodal cilia of wild-type mice showed rotary movement (Fig. 4f); however, Efcab1 −/− nodal cilia moved in an irregular, not rotary manner and often in planar trajectories (Fig. 4g). To check if this aberrant ciliary motility and fluid flow resulted in situs inversus, we examined the gene expression of Nodal and Lefty, both of which are expressed in the left plate mesoderm (LPM) and are key genes in the determination of left-right asymmetry 35 . Both genes were expressed in the LPM of wild-type E8.0 embryos but in Efcab1 −/− embryos both genes were expressed on both sides (Fig. 4h).
Since two populations of cilia are known in mouse node, we examined the localization of Efcab1 by immunofluorescent staining. In wild-type embryos, the cilia in the central region of node were recognized with anti-Efcab1 antibody but no significant staining was observed in cilia of the peripheral region ( Fig. 4i). As demonstrated by scanning electron microscopy ( Fig. 4a, b), we could not detect cilia of the central node region in Efcab1 −/− mice. However, significant numbers of cilia were observed in the peripheral region of these mice (Fig. 4i).
Calaxin-deficient zebrafish show situs inversus but normal cilia formation in Kupffer's vesicle. To investigate the conservation of vertebrate calaxin function in the determination of body laterality, we carried out CRISPR/Cas9-mediated knockout of efcab1 in zebrafish. We identified an 11 bp-insertion in exon 2 of efcab1, including an in-frame stop codon ( Fig. 5a-c), which resulted in the loss of Efcab1 protein expression (Fig. 5d, Supplementary Fig. 4). In zebrafish, cilia-directed flow in Kupffer's vesicle (KV) has a critical role in establishing the left-right body axis 36 . In contrast to the drastic loss of nodal cilia formation in Efcab1 −/− mouse embryos, efcab1 −/− zebrafish showed normal formation of KV cilia (Fig. 5e); no significant difference was observed in the number of cilia between wild-type and efcab1 −/− fish (Fig. 5f). However, many of the KV cilia in efcab1 −/− zebrafish beat with an irregular cycle ( . The abnormal ciliary movements in efcab1 −/− fish induced laterality defects in embryos. In wild-type embryos, the heart ventricle loops toward the right and the atrium loops toward the left. However, in almost half of the efcab1 −/− embryos the direction of the heart loop was reversed. The reversed loop was rescued by injection of efcab1 mRNA (Fig. 5j, k).

Discussion
Knockout of mouse calaxin, a Ca 2+ sensor for ODAs, caused several defects commonly seen in PCD. Despite apparent normal formation of epithelial cilia and sperm flagella, calaxin deficiency resulted in striking inhibition of nodal cilia formation, which could be one reason for the loss of nodal flow. However, induction of left side-specific gene expression requires at least two nodal cilia 37 . Therefore, although Efcab1 −/− mice have a greatly reduced number of nodal cilia (~5 cilia on average) this cannot explain why half showed situs inversus. Rather, we consider aberrant ciliary motility with planar beating, not smooth rotational movement, to be critical for the laterality defect. The common feature of structural normality of 9 + 2 cilia between calaxin knockout mice and PCD patients with DNAH11 (β-type HC of the ODA 24 ) mutations suggests a possible relationship between these two proteins.
Calaxin plays an important role in propagation of the asymmetric waveform in ascidian sperm flagella 30 . This is consistent with our results showing suppression of propagation of the pro-hook bend in the sperm flagella of calaxin-deficient mice. However, calaxin-deficient mice are fertile, indicating that calaxindependent regulation of the flagellar bend propagation might not be essential for successful fertilization, at least in mice, and that sperm with abnormal propagation of the pro-hook bend can penetrate the zona pellucida to achieve fertilization. This is in contrast to mutant mice lacking sperm-specific calcineurin (PPP3C/PPP3R2), which show defects in bend formation in the proximal region of the flagellum and are defective in zona penetration 38 .
Loss of calaxin resulted in a narrower spatial range of beating in the trachea and brain multicilia of mice. A quite similar effect is also observed in the monocilia of calaxin-deficient sea urchin embryos 32 , suggesting that a major role of calaxin in the regulation of ciliary beating is likely common between mice and sea urchins. In sea urchin embryos, any deficiency of calaxin causes disruption to their basal body orientation, leading to uncoordinated ciliary movement and aberrant swimming 32 . However, despite clear changes in the ciliary waveform and beating of each cilium, the effects of calaxin-knockout on the epithelial fluid flow in mouse multicilia were not overly drastic. It is likely that weak beating of each cilium would not affect the ciliary orientation in multicilia, such as observed in sea urchin monocilia, as it has been suggested that the mechanical feedback of hydrodynamic force within multicilia would compensate for this [39][40][41] . Knockout of calaxin caused reduced motility in both the sperm flagella and epithelial multicilia, without any apparent changes in the 9 + 2 structure. In rare cases, ODAs on the doublet 3, 5, and 8 in sperm flagella and the doublet 8 and 9 in trachea cilia were lost, suggesting that calaxin might play some role in the stability of ODAs, particularly for those on the doublets located either side of the plane of central pair complex which are strongly involved for flagellar or ciliary bending 27 . In contrast to the apparently normal formation of sperm flagella and epithelial multicilia, calaxin mutants were deficient in the formation of nodal cilia, indicating a distinct mechanism for ciliary formation in the node.
Why is calaxin essential for the formation of nodal cilia only? It could be related to the type of cilia, i.e., mono-cilia or multi-cilia; however, knockdown of calaxin in sea urchin embryos results in no structural changes in ectoderm monocilia 32 . Moreover, knockout of calaxin in zebrafish resulted in normal formation of KV monocilia. The most probable explanation for the specific effect on ciliary formation is that mouse node cilia have a 9 + 0 axonemal structure, unlike the 9 + 2 structure in other motile cilia. There have been conflicting reports as to whether the dynein species present in nodal cilia possess ODAs and/or IDAs; however, they most probably only possess ODAs 10,42 . When ODAs are assembled in a 9 + 0 arrangement, calaxin might have a critical role in ciliogenesis. This could also indicate a novel mechanism where the Ca 2+ -dependent dynein sensor, calaxin, is involved in intraflagellar transport, docking axonemal components at the basal body/transition zone, or maintenance of axonemal integrity. Although intraflagellar transport is regulated by intracellular Ca 2+ concentration 43 , the Ca 2+ -binding proteins that respond to the changes in intracellular [Ca 2+ ] are yet to be identified. It is possible that Ca 2+ -binding axonemal proteins 3,27,44,45 , such as ODA calaxin, IDA centrin, radial spoke calmodulin (CaM), and CaM-binding proteins in radial spoke/ central apparatus, might be involved in the regulation of intraflagellar transport. This idea is consistent with zebrafish 9 + 2 KV cilia with ODAs, IDAs, radial spokes and central apparatus, being normally formed in calaxin-knockout embryos, although motility was sufficiently altered to induce a laterality defect. Two populations of nodal cilia, including those in central region with 9 + 0 and in peripheral region with 9 + 2 structures, have been observed in the mouse node 46,47 . This is consistent with the fact that LRD is only localized in the cilia of central node region 46 and suggests that calaxin is closely related to the function of LRD motor activity.
Asymmetry of intracellular Ca 2+ dynamics in the node has been observed 48,49 but Ca 2+ -dependent regulation of nodal cilia motility is less understood. Attempts to image Ca 2+ dynamics in the node have produced different conclusions regarding the necessity of Ca 2+ dynamics in left-right asymmetry 49,50 . No change in motility of KV cilia upon changes in intracellular Ca 2+ has been observed in zebrafish embryos 51 . However, it is not known how intracellular Ca 2+ regulates ciliary waveforms. We found that knockout of calaxin causes the change from rotatory movement with proper ciliary bending to planar movement with less bending, indicating that nodal ciliary motility depends on intracellular Ca 2+ -dependent regulation of calaxin. Ca 2+ -binding to calaxin appears to occur at the second EF-hand between concentrations of 10 −7 to 10 −6 M intracellular [Ca 2+ ], which induces proper propagation of asymmetric flagellar bend in Ciona sperm 30,31 . The resting [Ca 2+ ] in a mouse node is reported to bẽ 300 nM 50 , which would represent a threshold for Ca 2+ -binding to calaxin. Therefore, we suggest that Ca 2+ would be bound to calaxin in the node and would regulate proper ciliary bending, resulting in the generation of left-ward fluid flow.
For zebrafish, embryos were fixed with 4% paraformaldehyde (PFA) in PBS, and then stored in methanol at −20°C. After rehydration with PBST, hybridization was performed overnight at 63°C with digoxigenin-labeled RNA probes. Hybridized specimens were washed with SSC (saline sodium citrate) buffer, then treated with AP-conjugated anti-digoxigenin Fab fragments (1:4000 dilution; Roche) in PBST at 4°C overnight. After washing with PBST, signals were developed using BM-purple (Roche). When desired intensities of staining were obtained, reactions were stopped. Before observation, specimens were transferred to 80% glycerol/PBS to make them transparent. Images were taken using a stereoscopic microscope (MVX10; Olympus) and a CCD camera (DP73; Olympus).
In vitro fertilization. Follicular development and ovulation were induced by hCG injection of 8-12-week-old female mice. Mature oocytes were recovered 14-16 h after the hormone treatment. Mature spermatozoa were obtained from the cauda epididymis of 8-10-week-old male mice. Spermatozoa were suspended and preincubated in TYH culture medium for 2 h and added to the oocyte suspension at 1.5 × 10 5 sperm/ml. After incubation at 37°C for 6 h, male and female pronuclei were stained with Hoechst 33342 and counted under a fluorescence microscope.
Transmission electron microscopy. Samples were fixed in 2.5% glutaraldehyde, 5 mM MgSO 4 , 0.1 M cacodylate buffer, pH 7.4, for 1 h at room temperature. After washing with 0.1 M cacodylate buffer, pH 7.4, the samples were post-fixed in 1% OsO 4 on ice for 1 h, dehydrated in a 30-100% EtOH series, substituted by propylene oxide and embedded in Epon 812 or Agar Low Viscosity Resin (LV-Resin). Sections (70 nm) were cut using an ultramicrotome (LEICA, Wetzlar, Germany) and mounted on neoprene-coated grids. Samples were stained with 7% uranyl acetate for 20 min, followed by Reynolds lead staining for 2 min, and observed under a transmission electron microscope (JEM 1200EX, JEOL).
Scanning electron microscopy (SEM). Samples were fixed and post-fixed by the same procedure used for transmission electron microscopy, dehydrated in a 30-100% EtOH series, substituted by t-butanol and frozen at −30°C. Samples were then lyophilized in a freeze-drying device (JFD-320, JEOL), mounted on an aluminum block and sputter coated with Au-Pd using an Auto Fine Coater JEC-3000FC (JEOL). Samples were observed under a SEM (JCM5000, JEOL).
Recording of mouse sperm flagellar motility. Spermatozoa were collected from the cauda epididymis, suspended in TYH culture medium and incubated under mineral oil (Nacalai Tesque) at 37°C in 5% CO 2 . Sperm motility was observed in a warm chamber (Leja, 20 μm depth, NeuroScience Osceola, WI) with a 10× objective on a BX51 phase contrast microscope (Olympus) and recorded at 500 frames per second (fps) through a HAS-D3 high-speed camera (DITECT, Japan). Velocity of sperm swimming was calculated from the trajectory for 0.5 s.
Recording of cilia motility in mouse trachea and brain. Tracheal cilia were observed according to a previously described method 55 . Mouse trachea were removed by dissection and placed in DH10 culture medium containing 8.3 g/l DMEM powder (Sigma D5030), 25 mM HEPES-NaOH, pH 7.2, 4.5 g/l glucose, 0.11 g/l sodium pyruvate, 10% FBS. Trachea were opened on the dorsal side and cut into 3 mm squares under a stereoscopic microscope. The tissue pieces were transferred into medium containing 5 mM DTT and incubated for 5 min. The tissue pieces were then transferred to DH10, rinsed for 5 min twice, and observed in a chamber on a glass slide (26 mm × 76 mm, Toshin Riko) with a Scotch tape spacer under a 40× or 100× objective (UPlan FL, Olympus) on a differential interference contrast microscope (BX51, Olympus).
Ependymal cilia were observed according to Ibanez-Tallon et al. 56 . Brains were dissected from 13 to 17-week-old mice, transferred to DH10 culture medium, and trimmed until the epithelium of the lateral ventricle became exposed. Tissue slices of 150 µm thickness were prepared using a Linear Slicer PRO7 (D.S.K Co., Japan) and mounted in medium in a chamber on a glass slide (S1111, Matsunami). The chamber was made by placing a plastic tape as a spacer. The silicon sheet prevented compression of the tissue after sealing the hole with a coverslip. The waveform of ependymal cilia was observed under a 60× objective (UPlan FL, Olympus) on a phase contrast microscope (BX51, Olympus) and recorded at 200 fps through a HAS-220 high-speed camera (DITECT).
Recording of nodal cilia. Nodal cilia were observed according to Nonaka et al. 57 . Embryos were isolated from decidua by removal of Reichert's membrane in DH10 medium. Tissue fragments containing the node were excised using a pair of needles and transferred into medium containing 8.3 g/l DMEM powder (Sigma D5030), 25 mM HEPES-NaOH (pH 7.2), 4.5 g/l glucose, 0.11 g/l sodium pyruvate, and 50% rat IC serum 57 . A 0.3 mm thick silicon sheet with a hole was set onto a glass slide and a node fragment was transferred in medium with the nodal cilia face up. The specimen was then covered with a coverslip and observed under a differential interference contrast microscope (BX51, Olympus) with a 40× or 60× objective.
Analysis of ciliary movement and fluid flow. Fluid flow was traced using 1 µm fluorescent beads (FluoSphere, F-8820, Invitrogen) and recorded at 200 fps by a fluorescence microscope (BX51, Olympus) equipped with a high-speed camera HAS-220 (DITECT) and a 20× objective for tracheal and ependymal cilia or a 60× objective for nodal cilia. Flagellar and ciliary motility and the trajectories of beads were analyzed by motility analysis software, Bohboh 58 . To analyze the trajectories of tracheal ciliary beating, fluorescent beads attached near the tips of cilia were recorded and traced by Bohboh. For nodal cilia, recorded images were processed to reduce the background and the positions of cilium tips were traced by Bohboh.
Kupffer's vesicle cilia analysis. Embryos with a Kupffer's vesicle were selected at 12 hpf and dechrionated before observation. For orientation, embryos were embedded in 0.8% low gelling temperature agarose (Sigma) in 1/3 Ringer's solution. Motility of Kupffer's vesicle cilia was observed under bright-field conditions using an inverted microscope (DMI6000B; Leica) and a high-speed camera (HAS-L1; Detect) at 1000 fps.
Antibodies, immunoblotting, and immunofluorescence microscopy. A cDNA fragment encoding mouse Efcab1 was PCR-amplified, subcloned into a pET-23d vector (Novagen) and expressed in E. coli BL21 (DE3). The recombinant protein produced was then purified using a His-tag affinity column. A polyclonal antibody against mouse Efcab1 was produced in 16 rabbits via immunization with the recombinant protein. For immunoblot analysis, proteins were extracted by 4 M urea, 1% CHAPS, 20 mM HEPES-NaOH (pH 7.5) or directly by the SDS-buffer for epithelial tissues or sperm, respectively, then separated by SDS-polyacrylamide gel electrophoresis and transferred to a polyvinylidene difluoride membrane. The membranes were blocked by PBST (PBS containing 0.1% Tween 20) containing 7.5% skimmed milk, followed by incubation with the anti-mouse Efcab1 antibody (1:2000) or by a mouse monoclonal antibody against tubulin-α Ab-2 (Clone DM1A) (Thermo Fisher Scientific, 1:10,000). Blots were then incubated with HRPconjugated secondary antibodies (1:10,000), washed with PBST and developed with ECL-Prime enhanced chemiluminescence substrate kit (GE Healthcares).
Mature spermatozoa were collected from the cauda epididymis, suspended in phosphate-buffered saline (PBS) and immobilized on a poly-lysine coated glass slide. They were fixed in cold methanol (−20°C), dehydrated with PBS, permeabilized with T-PBS (0.1% Triton X-100 in PBS) and blocked with 10% goat serum in T-PBS for 2 h. After blocking, samples were incubated in the blocking buffer containing a rabbit polyclonal antibody against mouse Efcab1 at 1:100 dilution for 1 h. After washing with T-PBS three times for 1 h, samples were treated with secondary antibodies (Alexa FluorR 488-labeled secondary antibody against rabbit IgG, Invitrogen) at 1:1000 dilution and β-Tubulin-Cy3 (C4585, Sigma-Aldrich) at 1:100 dilution for 1 h. For trachea and brain, tissues were fixed in 4% PFA in PBS for 2 h at 4°C and then stored in methanol at −20°C. After rehydration with PBSDT (1% DMSO, 0.1% TritonX-100 in PBS), samples were treated with a blocking buffer containing 10% goat serum in T-PBS. After blocking, samples were incubated in the blocking buffer containing a rabbit polyclonal antibody against mouse calaxin at 1:200 dilution as well as a mouse monoclonal antibody against acetylated α-tubulin (D20G3, Cell Signaling Technology) at 1:400 dilution for 2 h. After washing with T-PBS three times over 1 h, samples were treated with secondary antibodies (Alexa Fluor® 488-labeled secondary antibody against rabbit IgG, Invitrogen, and Alexa Fluor® 546-labeled secondary antibody against mouse IgG, Invitrogen) at 1:200 dilution for 1 h. Samples were washed by T-PBS three times over 1 h, followed by incubation in PBS. For nodal cilia, embryos devoid of Reichert's membrane were fixed in 4% PFA in PBS for 30 min at 4°C, washed with T-PBS and treated with cold methanol for 20 min. Subsequent procedures were the same as those employed for trachea and brain, except that antibodies were diluted in the Solution B of Can Get Signal Immunostain (TOYOBO). In trachea and brain samples, 4′,6-diamidino-2-phenylindole (DAPI) was added to PBS at 1 μM before mounting on a glass slide. Observations were made using a fluorescence microscope (Olympus BX53) with a digital camera (DP74, Olympus) for sperm and a confocal microscopy (Fluoview FV10i, Olympus) for epithelial tissues.
Data accession. Other relevant information regarding data accession is described in the Supplementary information (Supplementary Note 1).
Reporting summary. Further information on research design is available in the Nature Research Reporting Summary linked to this article.

Data availability
Movies associated with the current study are available online at https://doi.org/10.6084/ m9.figshare.8121296 59 . All other data generated during and/or analyzed during the currently study are available from the corresponding author on reasonable request.