Abstract
The establishment of Left/Right (LR) asymmetry is fundamental to animal development. While the pathways governing antero-posterior and dorso-ventral patterning are well conserved among different phyla, divergent mechanisms have been implicated in the specification of LR asymmetry in vertebrates and invertebrates. A cilia-driven, directional fluid flow is important for symmetry breaking in numerous vertebrates, including zebrafish1–⇓⇓⇓⇓⇓⇓⇓⇓10. Alternatively, LR asymmetry can be established independently of motile cilia, notably through the intrinsic chirality of the acto-myosin cytoskeleton11–⇓⇓⇓⇓⇓⇓18. Here we show that MyosiniD (Myo1D), which has been previously identified as a key determinant of LR asymmetry in Drosophila12,13, is essential for the formation and the function of the zebrafish LR Organizer (LRO). We show that Myo1D controls the polarity of LRO cilia and interacts functionally with the Planar Cell Polarity (PCP) gene VanGogh-like2 (Vangl2)19, to promote the establishment of a functional LRO flow. Our findings identify Myo1D as the first evolutionarily conserved determinant of LR asymmetry, and show that functional interactions between Myo1D and PCP are central to the establishment of animal LR asymmetry.
Numerous studies have revealed an intriguing diversity in the strategies establishing laterality both within and across phyla, making the identification of a unifying mechanism elusive14,20,21. An attractive hypothesis however, is that the chirality of actin filaments may provide a conserved template for organismal LR asymmetry14,18. Accordingly, actin-binding proteins govern molecular22 and cellular chirality23 and actin-dependent processes regulate invertebrate and vertebrate laterality12–⇓14,17,24. Studies in Drosophila have identified the actin-based molecular motor protein Myosin1D (Myo1D, a.k.a. Myosin31DF) as an essential determinant of LR asymmetry12,13,25. Here, we study the contribution of this conserved factor to vertebrate laterality, by analyzing its function in zebrafish.
We first determined whether zebrafish and Drosophila Myo1D have conserved activities by testing the ability of fish myo1D to rescue the laterality defects of fly myo1D mutants. The two orthologous proteins show 69% sequence similarity. Remarkably, expressing zebrafish myo1D in Drosophila fully restored the chirality of genitalia rotation (Fig. 1a), a prominent LR marker in Drosophila12,13. These data indicate that zebrafish Myo1D is capable of mediating a conserved LR symmetry-breaking activity.
To study zebrafish myo1D function, we generated frameshift mutations that disrupt the P-loop required for Myo1D motor activity26 and that delete the actin- and cargo-binding domains (Extended Data Fig. 1a). Homozygous mutants lacking zygotic myo1D function develop normally and give rise to fertile adults (Extended Data Fig. 1b). In situ hybridization reveals that, in addition to their zygotic expression, myo1D transcripts are maternally provided (Extended Data Fig. 1c-e). We therefore used homozygous mutant females to generate embryos lacking either only maternal (M) or both maternal and zygotic (MZ) myo1D activities. M and MZ myo1D mutants display LR asymmetry defects at the level of the heart (Fig. 1b-i), viscera (Fig. 1j-m) and brain (Extended Data Fig. 2a-f). Identical phenotypes were observed using two different mutant alleles (Fig. 1e,i). Similar defects are moreover generated when a morpholino that blocks the translation of both maternally deposited and zygotically transcribed myo1D RNA is injected in wild-type embryos (Extended Data Fig. 2g-m). Altogether, our experiments identify Myo1D as a conserved regulator of laterality in both vertebrates and invertebrates.
In Drosophila, Myosin1C (Myo1C, a.k.a. Myo61F) acts as an antagonist of the essential dextral determinant Myo1D27. In flies, myo1C overexpression phenocopies myo1D loss-of-function. However, Drosophila myo1C is itself dispensable for LR asymmetry27. The zebrafish genome encodes two myo1C homologues, myosinlCa and myosinlCb (myo1Ca & b). Like in Drosophila, zygotic myo1Ca mutants and MZ myo1Cb mutants do not display LR defects (not shown). Interestingly, Myo1Cb overexpression using mRNA microinjection causes cardiac laterality defects (Fig. 1n,o), thus mimicking myo1D loss-of-function, as observed in Drosophila27.
We noticed that the overexpression of zebrafish myo1Cb enhances MZ myo1D mutant LR defects (Fig. 1n-o), raising the question whether additional Myo1D-like activities might contribute to the regulation of LR asymmetry. Genome analysis shows that the zebrafish Myosin1G (Myo1G) protein is closely related to Myo1D (79% similar amino acids). While myo1G knock-down through morpholino or Crispr/Cas9 nuclease injection elicits modest cardiac laterality defects (Fig. 1n,o), the injection of these reagents into MZ myo1D mutant embryos allows to increase the frequency of LR defects to a level equivalent to the one observed upon myo1Cb overexpression (Fig. 1n,o). Altogether, these findings provide evidence that Myo1D-like agonists (Myo1D & G) and their antagonist Myo1Cb form an evolutionarily conserved pathway for the control of LR asymmetry.
The observation that MZ myo1D mutants present laterality defects at the level of the heart, viscera and brain (Fig. 1b-m, Extended Data Fig. 2a-f), suggests that myo1D may control the formation and/or function of the central zebrafish LR Organizer (LRO), Kupffer’s Vesicle (KV)3,28. The ciliary LRO flow induces the expression of both the Nodal ligand southpaw (spaw) and the transcription factor pitx2 in the left lateral plate mesodem29,30. Consistent with a potential role of Myo1D at the level of the LRO, the asymmetric expression of these two genes is disrupted in MZ myo1D mutants (Fig. 2a-f). To further test Myo1D’s function in the LRO, we looked at the asymmetric expression of the TGFβ antagonist charon/dand5, which is the earliest known transcriptional response to the KV flow4,31. While weak symmetric dand5 expression is initiated prior to the establishment of a functional KV, dand5 asymmetry arises as the LRO matures, peaks by the 8-somite stage and decreases again during later development (Fig. 2g-m). In accordance with a dysfunction of the LRO, dand5 asymmetry is reduced in MZ myo1D mutants and completely lost in MZ myo1D mutants injected with myo1Cb RNA or myo1G morpholino (Fig. 2n-t). These data demonstrate that prior to the first morphological manifestations of organ LR asymmetry, Myo1D is required at early stages for the LRO-dependent establishment of asymmetric gene expression.
To further dissect the role of myo1D in zebrafish LR asymmetry, we analyzed its contribution to the formation and function of the KV. While myo1D is dispensable for the migration and coalescence of sox17-positive KV precursor cells (Extended Data Fig. 3), we found that the average KV size and cilia number are reduced in MZ myo1D mutants (Fig. 3a,b,e,g-i). As LRO function requires a minimal organ size and number of cilia4,32, we monitored KV size, cilia number and laterality in individual embryos. No correlation could be observed between organ size, cilia number and the appearance of laterality defects (Fig. 3e, Extended Data Fig. 4a). Accordingly, treatment with IBMX and Forskolin, which promote KV lumen inflation33, rescues KV size but not laterality defects in MZ myo1D mutants (Fig. 3c-f). While the loss of myo1D causes a modest reduction of cilia size (Fig. 3g,h,j), it has no incidence on their motility or their clustering in the anterior KV half, two factors known to be required for LRO flow (Extended Data Fig. 4b,c)34.
To visualize the LRO flow, we tracked fluorescent microspheres in the KV lumen. While wild-type controls mostly displayed circular trajectories (Fig. 3k), MZ myo1D mutants often presented aberrant flow patterns (Fig. 3l). To quantitatively assess the LRO flow, we determined its mean angular velocity as a measure of the effective circular flow strength (see Methods). Results show that MZ myo1D mutants display a decrease in angular flow velocity (Fig. 3m). Importantly, LR asymmetry defects in MZ myo1D mutants correlate with pronounced reductions in flow velocity (Fig. 3m). In contrast, treatment of control wild-type embryos with Ouabain to pharmacologically reduce KV size leads to an increase in flow velocity with no associated LR defects (Extended Data Fig. 5). These data show that Myo1D is required for the establishment of a functional LRO flow, independently of KV size.
How does Myo1D control KV flow? It is known that embryos with no, or immotile, cilia lack a detectable flow3,4. In MZ myo1D mutants, cilia are motile but the LRO flow is erratic, a phenotype reminiscent of mutants for the Planar Cell Polarity (PCP) pathway component vangl2 (Fig. 4a-c)19,35. We confirmed this phenotypic similarity through quantitative analysis of KV flow in vangl2m209 mutants (Fig. 4e,f). Previous work showed that the establishment of LR asymmetry requires the generation of peak flow velocities in the anterior KV4,34. We found that both MZ myo1D and vangl2 mutants display a reduction in overall angular velocity (Fig. 4e) and a disruption of the antero-posterior flow pattern (Fig. 4f). Interestingly, while myo1D and vangl2 mutants have similar flow phenotypes, removal of one copy of vangl2 reduces the penetrance of LR defects in M or MZ myo1D embryos (Fig. 4g), suggesting that the two genes affect the LRO in opposite ways.
PCP signaling controls the posterior positioning of the ciliary basal body and the orientation of the ciliary rotation cone to promote LRO flow19,36–⇓38. Recently, rat myo1D has been shown to control cilia orientation in multiciliated epithelia39. We therefore imaged cilia in live embryos and found that zebrafish myo1D and vangl2 both affect cilia orientation: While vangl2 mutants present an excess of anteriorly pointing cilia in the anterior KV, the loss of myo1D increases the number of posteriorly pointing cilia in the posterior organ half (Fig.4h-l, Extended Data Fig. 6). Additionally, the inactivation of either myo1D or vangl2 causes an anterior displacement of the basal body (Fig. 4m-p). The positioning of the basal body is however partially restored when one copy of vangl2 is removed in MZ myo1D mutants (Fig. 4m,q). These findings suggest that myo1D and vangl2 affect PCP-dependent KV cyto-architecture, but do so in opposite ways, consistent with our genetic interaction data (Fig. 4g). We therefore reasoned that if the observed cilia defects are relevant for LRO function, then a reduction of vangl2 gene dosage should restore KV flow in MZ myo1D mutants. Accordingly, the removal of one copy of vangl2 allows to partially rescue KV flow pattern and velocity in MZ myo1D mutants (Fig. 4d,e,f).
Taken together, our data show that Myo1D is essential for the formation and function of the zebrafish LRO, identifying the first common determinant of LR asymmetry in both vertebrates and invertebrates. Our work also reveals that a functional interaction between zebrafish Myo1D and PCP controls the positioning and orientation of KV cilia that are required for the establishment of a directional, symmetry-breaking flow (Fig. 4r,s). Interestingly, recent work has shown a functional link between Myo1D and PCP pathways in Drosophila40 and in Xenopus (M. Blum, personal communication). We propose that the coupling between Myo1D and PCP polarity systems represent an evolutionarily conserved genetic framework controlling LR asymmetry in animals.
AUTHOR CONTRIBUTIONS
The genetic analysis of MyosinI function in zebrafish LR asymmetry was entirely performed by T.J. J.B.C. performed initial morpholino knock-down experiments of zMyo1D and zMyo1G. S.P. cloned zMyo1D and G, analyzed Myo1D expression and performed Myo1D morpholino knock-down and Myo1Cb overexpression experiments together with T.J. D.C. generated zMyo1D-transgenic Drosophila lines. C.G. performed the rescue experiments of the fly Myo1D mutants with zMyo1D. S.N. and M.F. designed the study, analyzed the data and wrote the manuscript.
AUTHOR INFORMATION
The authors declare no competing financial interests. Correspondence and request for materials should be addressed to S.N. (noselli{at}unice.fr) or M.F. (furthauer{at}unice.fr).
METHODS
Fly strains and genetics
Flies were grown according to standard procedures. For inter-specific rescue experiments, the UAS::zMyo1D transgene (Zebrafish Myo1D ORF cloned into pUAST vector) was expressed either in myo1D depleted flies (w1118 ; ptc::GAL4, myo1Dk1 ; UAS::dmyo1D-RNAi 13) or in a myo1Dk213 null mutant background (da::GAL4; myo1Dk2)(da::GAL4, Bloomington Drosophila Stock Center #55851). Two independent insertions of the UAS::zMyo1D transgene (line 3 and 6) were used and showed similar results.
Rotation of the male genitalia was scored as previously described42. Briefly, male genital plates are scored according to the angle made between the dorso-ventral and anus-penis axes when viewed from the posterior. Full dextral/clockwise (+360°) corresponding to the wild-type situation; partial sinistral/counterclockwise rotation (from −1 to −359°); full sinistral/counterclockwise rotation (−360°).
Zebrafish strains and embryo maintenance
Embryo were raised in 0.3x Danieau medium (17.4 mM NaCl, 0.21 mM KCl, 0.12 mM MgSO4, 0.18 mM Ca(NO3)2, 1.5 mM Hepes, pH 7.6) at 28,5°C, and staged according to standard criteria43. If necessary, 1-phenyl-2-thiourea (Sigma) was added at 30 mg/l to prevent embryonic pigmentation.
Crispr/Cas9 Mutagenesis
Crispr/Cas9 mutagenesis of zebrafish myo1D (Ensembl gene ENSDARG00000036863) was performed in a wild-type TÜ background. gRNA design and in vitro transcription of Cas9 RNA were performed according to reported protocols44. gRNA was transcribed from a template oligo as previously described45. The sequence of the selected Crisp target site was 5’-GAGTGGAGCTGGAAAAACAGAGG-3’ (bold lettering indicates the PAM motif). The efficiency of Crispr/Cas9-induced mutagenesis was monitored at 24 hpf using a T7 endonuclease assay44 on a PCR amplicon comprising the Crispr target region (Forward primer: 5’-TCTTCACTGACACTGGTATG-3’, Reverse primer: 5’-CCATCACTGCAGCAGAAATGAGAG-3’). Adult F0 fish were outcrossed to AB wild-type fish and DNA extracted from F1 progeny. Mutations were identified through direct sequencing of the same PCR amplicon that was used in the T7 endonuclease assay.
Zebrafish myo1G (Ensembl gene ENSDARG00000036104) was inactivated through Crispr/Cas9 mutagenesis at the target site 5’-GATGTCATTGAGGACTACAGGGG-3’. A pre-assembled complex of purified Cas9 protein (NEB, 954 ng/μl) and myo1G gRNA (200 ng/μl) was injected into one cell stage embryos and the cutting efficiency monitored using a T7 endonuclease assay on a myo1G PCR amplicon (Forward primer: 5′-GGAGAAGTAGTGGTGTCCGTTAAC-3’, Reverse primer: 5’-CTCACTTTTGGGCTAACAGCTC-3’).
Fish strains and molecular genotyping
The following lines were used for this work: Tg(−5.0sox17:EGFP)zf9941, Tg(cmlc2:RFP)46, vangl2m209 47.
Wild-type and mutant alleles of myo1D were identified through allele-specific PCRs. The allele-specific forward primers 5’-TGGAGCTGGAAAAAGGCTCGT-3’ (myo1Dtj16b) and 5’-GTGGAGCTGGAAAAAGGCTATAC-3’ (myo1Dtj16c) were used together with the generic reverse primer 5’-CCATCACTGCAGCAGAAATGAGAG-3’ to genotype for the presence of different myo1D mutant alleles. PCR amplicon size was 133 bp for myo1Dtj16b and 145 bp for myo1Dtj16c. The wild-type myo1D allele was amplified using the forward primer 5’-AGAGTGGAGCTGGAAAAACAGA-3’ and the reverse primer 5’-CCCATCCCTCGTGTGAAACTAAATCAC-3’ to yield a 339 bp amplicon. PCR amplification with GoTagG2 polymerase (Promega) was performed using the following cycling conditions: Initial denaturation 2min 95°C; 10 cycles [30s 95°C, 30s 65°C first to 55°C last cycle, 30s 72°C]; 20 cycles [30s 95°C, 30s 55°C, 30s 72°C]; final extension 5min 72°C.
To generate maternal zygotic myo1D mutants (MZ myo1Dtj16b/tj16b or MZ myo1Dtj16c/tj16c), myo1D heterozygous fish were incrossed and the progeny raised to adulthood. PCR genotyping of these animals allowed identifying homozygous mutant as well as homozygous wild-type fish which were then used to produce MZ myo1Dtj16b/t16b or MZ myo1Dtj16c/tj16c mutant embryos, as well as wild-type myo1D+/+ control embryos. myo1Dtj16b or myo1Dtj16c homozygous mutant animals presented similar laterality defects (Fig. 1e,i). Unless mentioned otherwise, experiments were performed using the myo1Dtj16b allele.
For the genetic interaction experiment displayed in Fig. 4g, myo1Dtj16b homozygous mutant females were crossed with myo1Dtj16b/+; vangl2m209/+ males. A total of 912 embryos derived from 6 independent crosses were scored and molecular genotyping performed for all embryos presenting cardiac laterality defects. Genotyping for the vangl2m209 mutant allele was performed as previously described48.
Plasmid generation
The myo1Cb ORF was amplified from a mixed stage pool of cDNAs using the primers: 5’-GATCCCATCGATTCGACGGAATCCGGGTCATGATG-3’ and 5’-AGGCTCGAGAGGCCTTGGTCACGACTCGTCCATC-3’ and cloned into the pCS2+ vector. Bold letterings indicate primer overhangs that are also present in the pCS2+ sequence and were used to recombine insert and vector using Gibson assembly (NEB). A similar strategy was used to clone the myo1D ORF into pCS2+ using the primers 5’-GATCCCATCGATTCGACAGCTTATTATGGCAGAACACG-3’ and 5’-AGGCTCGAGAGGCCTTAAATGGGATCCTGGTCCTCTAG-3’.
RNA and morpholino injections
myo1Cb-pCS2+ was linearized with KpnI, myo1Cb RNA in vitro synthesized using the mMessage mMachine SP6 transcription kit (Ambion) and the resulting RNA injected at 50ng/μl. RNA encoding Arl13b-GFP was made as previously described19 and injected at 20ng/μl. Two translations-blocking morpholinos (Gene Tools) were used: Myo1D, 5’-ACTTTCGTGTTCTGCCATAATAAGC-3’ (injected at 750 μM); Myo1G, 5’-CCTCCAGCTCCGCCATCCCTGCAAC-3’ (62.5 μM). All reagents were injected at the one cell stage at a volume of 0.88nl together with 0.2% phenol red.
RNA in situ hybridizations
Whole mount RNA in situ hybridizations were performed as previously described49. The following probes have been previously described: cmlc250, foxa1/fkd751, lefty152, pitx230, sox1753, southpaw29, dand5/charon31. To generate a myo1D probe, myo1D was amplified from myo1D-pCS2+ using the following primers 5’-CGCTAATACGACTCACTATAGGGAGAGTTCTAGAGGCTCGAGAGG-3’ and 5’-GGCAAGACCAAAGTTTTCATCCGC-3’. The in situ probe was then synthesized directly from the PCR product using the incorporated T7 promoter (bold lettering). In situ hybridizations were documented using a Leica M205FA stereomicroscope equipped with an Infinity2-5C color CCD (Lumenera).
Immunohistochemistry
Dechorionated embryos were fixed overnight at 4°C in PEM (80 mM Sodium-Pipes, 5 mM EGTA, 1 mM MgCl2) − 4% PFA − 0.2% TritonX100. After washing 2 × 5 min in PEMT (PEM - 0.2% TritonX100), 10 min in PEM - 50 mM NH4Cl, 2 × 5 min in PEMT and blocking in PEMT - 2%BSA, embryos were incubated 2 hrs at room temperature with primary antibodies. Following incubation, embryos were washed during 5, 10, 15 and 20 min in PEMT, blocked in PEMT - 2%BSA, and incubated again with secondary antibodies for 2 hrs. Embryos were again washed during 5, 10, 15 and 20 min in PEMT, mounted in PEM-0,75% LMP-Agarose and analyzed on a Zeiss LSM710 laser scanning confocal microscope. The following primary antibodies were used in this study: Mouse@ZO-1 (1:250, Invitrogen 33-9100), Mouse@Acetylated Tubulin (1:2000, Sigma T7451) and Rabbit@yTubulin (1:250, Sigma T5192). The secondary antibodies Goat@MouseIgG1-Alexa568, Goat@MouseIgG2b-Alexa647 and Goat@Rabbit-Alexa488 were all obtained from Invitrogen and used at a dilution of 1:500.
Pharmacological treatments
Stock solutions of IBMX (100mM, Sigma) and Forskolin (10mM, Sigma) were prepared in DMSO. Ouabain (1mM, Sigma) was prepared in distilled water. Embryos were treated with 10 μM Forskolin and 40 μM IBMX, or 5 μM Ouabain at bud stage and washed in Danieau medium after imaging at the 8-somite stage.
Microinjection and tracking of fluorescent beads in the Left/Right Organizer
Fluoresbrite Polychromatic Red Microspheres 0.5μm (Polysciences) were manually injected into the KV lumen. The movement of fluorescent beads was imaged on a wide field Zeiss Axio Observer Z1 microscope with a 40X EC Plan-Neofluar PH/DCII NA 0.75 dry objective using an Andor Neo sCMOS camera and a DsRed filter. In this setup we obtained images with a pixel width of 0.1625 μm and recorded the KV flow for 2 minutes with a frame interval of 0.04s. Automated tracking of fluorescent beads was performed using a custom-made ImageJ script (available on request). The macro generates an automated tracking file that is compatible with the ImageJ MTrackJ plugin. Following automated detection, tracks were manually curated using MTrackJ to correct for any errors of the tracking algorithm and the tracking parameters exported in a .CSV datasheet for further analysis. Using this procedure, an average of 41624 tracking points were obtained per embryo. KV images were oriented precisely along the antero-posterior axis based on the position of the notochord as well as the characteristic indentation that is found at the anterior pole of the KV in the region were cells are most closely packed54. The center of the KV was determined as the point lying midway between the anterior and the posterior extremity of the KV on a line that extends posteriorly from the anterior pole of the KV. The angular flow velocities displayed in Figures 3 and 4 were calculated using the center of the KV as the point of origin of a polar coordinate system.
Quantification of ciliary orientation
KV motile cilia were labeled using Arl13b-GFP RNA microinjection and imaged on a Zeiss710 laser scanning confocal microscope. The entire KV was scanned with z-sections of 0,2 μm distance and a time interval of 0.48s between two consecutive planes. Movies were then treated in ImageJ with a Mean 3D filter (x radius=1, y radius=1, z radius=10) to visualize the ciliary rotation cone and be able to orient it from base to tip (center of the cone). Each cilium was drawn on the KV using the ImageJ arrow tool. Cilia positions and angles were then extracted from the arrow ROIs using a custom-made ImageJ script. To compare the position and orientation of cilia in different embryos as shown in Fig. 4h-j, the relative positions of different cilia were normalized in the common space of a circular virtual KV. Each cilium was placed on the virtual KV according to its relative radial position with respect to the organ border. ImageJ scripts are available on request.
Quantification of the antero-posterior positioning of the ciliary basal body
Confocal Z-stacks of the entire KV were acquired with a distance of 0,6 μm between two consecutive planes. The KV was oriented along the antero-posterior axis based on the position of the notochord. Basal bodies were identified as γ-Tubulin positive spots that are located at the base of an Acetylated Tubulin positive ciliary axoneme. The junctional marker ZO-1 was used to outline KV cells and determine the antero-posterior extent of each cell. We restrained our analysis to cells for which the entire outline could be unambiguously identified in the ZO-1 channel. As the KV is an ellipsoid structure, this applies essentially to the cells of the dorsal roof and ventral floor of the KV. Based on this criterium, we were able to determine the antero-posterior position of an average of 19.4 basal bodies per embryo. The anteroposterior location of the basal body was quantified by calculating the ratio between the distance of the basal body from the anterior-most limit of the ZO-1 signal and the full antero-posterior length of a cell inferred from the ZO-1 signal. According to this convention, a value of 0 would indicate a basal body that is located at the anterior extremity of the cell, while a value of 1 would correspond to an extreme posterior location.
dand5/charon quantification
dand5/charon in situ hybridization pictures were analyzed with ImageJ. A bounding rectangle was fit on the dand5 expression domain and the mean grey levels of the left and right part of the rectangle measured. Dand5 Right/Left enrichment was determined as the ratio between right and left mean grey levels as previously described55.
ACKNOWLEDGEMENTS
The present project was supported by the ANR project grants DRO-ASY (ANR-13-BSV2-006, S.N.) and DroZeMyo (M.F., S.N.). The work in the lab of M.F. was supported by a CNRS/INSERM ATIP/Avenir 2010 grant and an HFSP Career Development Award (00036/2010). T.J. benefited from a PhD fellowship from INSERM/Région PACA/Cancéropôle PACA and the Labex SIGNALIFE (ANR-11-LABX-0028-01). We thank S.Lopes, B.Ciruna and T.Lepage for the sharing of reagents and R.Rebillard for excellent fish care. Confocal microscopy and timelapse imaging of the KV fluid flow were performed at the iBV PRISM imaging platform with the help of Magali Mondin.