Cilia in vertebrate left–right patterning

Abstract

Understanding how left–right (LR) asymmetry is generated in vertebrate embryos is an important problem in developmental biology. In humans, a failure to align the left and right sides of cardiovascular and/or gastrointestinal systems often results in birth defects. Evidence from patients and animal models has implicated cilia in the process of left–right patterning. Here, we review the proposed functions for cilia in establishing LR asymmetry, which include creating transient leftward fluid flows in an embryonic ‘left–right organizer’. These flows direct asymmetric activation of a conserved Nodal (TGFβ) signalling pathway that guides asymmetric morphogenesis of developing organs. We discuss the leading hypotheses for how cilia-generated asymmetric fluid flows are translated into asymmetric molecular signals. We also discuss emerging mechanisms that control the subcellular positioning of cilia and the cellular architecture of the left–right organizer, both of which are critical for effective cilia function during left–right patterning. Finally, using mosaic cell-labelling and time-lapse imaging in the zebrafish embryo, we provide new evidence that precursor cells maintain their relative positions as they give rise to the ciliated left–right organizer. This suggests the possibility that these cells acquire left–right positional information prior to the appearance of cilia.

This article is part of the themed issue ‘Provocative questions in left–right asymmetry’.

1. Introduction

Left–right (LR) asymmetry, or laterality, is a conserved feature of the internal vertebrate body plan. During vertebrate embryogenesis, the primitive heart and gut tubes undergo asymmetric looping to give rise to distinct left and right sides and positions of internal organs. For example, the human heart, stomach and spleen are positioned on the left side of the body, whereas the liver lies on the right side. A perfect mirror-image inversion of this arrangement is called situs inversus totalis, but this carries little or no clinical consequence [1]. By contrast, severe birth defects can arise when laterality is perturbed in specific organs. This condition is referred to as heterotaxy syndrome and causes a broad spectrum of congenital malformations of the heart and gastrointestinal tract [2–4]. Studies in animal models have identified several conserved signalling pathways involved in LR patterning of the vertebrate embryo, but how asymmetric signals are initiated and then relayed to developing organs remains poorly understood.

Pioneering work in chick [5] and mouse [6] embryos in the 1990s identified molecular asymmetries that provide LR patterning information in vertebrate embryos. Signalling molecules were first found to be asymmetrically expressed around a transient embryonic organizer structure called the node (or Hensen's node). Expression of Nodal, a secreted TGF-β family ligand, subsequently expands into lateral plate mesoderm exclusively on the left side of the embryo to activate its own expression and its feedback inhibitor Lefty [7] and the homeobox transcription factor Pitx2 [8]. Unlike Nodal, asymmetric expression of Pitx2 persists in the developing heart and gut tubes to control asymmetric organ morphogenesis [9–11]. This asymmetric Nodal-Lefty-Pitx2 cassette is highly conserved in vertebrate embryos, and functional studies have demonstrated the importance of this signalling cascade for establishing organ laterality [12,13]. However, the mechanism(s) underlying the unilateral expression of this cascade remains a subject of debate.

Studies of a rare genetic disorder called Kartagener syndrome (KS) first provided an intriguing link between LR patterning and motile cilia. It was discovered that KS, which is characterized by respiratory dysfunction and male infertility, was caused by loss of cilia motility [14,15]. Motile cilia—microtubule-based whip-like projections from cells—beat in airways to drive mucociliary clearance of debris and pathogens; thus, respiratory defects are consistent with loss of cilia movement. Likewise, loss of motility of flagella, which are cilia-related structures that power sperm swimming, explains infertility in KS. Along with mucociliary and sperm problems, approximately 50% of KS patients have situs inversus totalis. This implicates motile cilia in establishing normal LR asymmetry of the body. In addition to situs inversus totalis, heterotaxy has also been found in primary ciliary dyskinesia (PCD) patients with cilia motility defects [16–18], indicating that loss of cilia motility can result in either reversed or ambiguous LR asymmetry [4]. Indeed, mutations in several genes involved in building ciliary ultrastructure have been identified in patients with laterality defects [19,20].

2. Functions of cilia in left–right patterning

(a). Motile cilia generate asymmetric fluid flows

The use of animal models has been instrumental for uncovering roles for cilia in LR development. A landmark study using knockout mice deficient for a kinesin protein first identified a transient group of monociliated cells in the ventral node of the embryo that are essential for establishing LR asymmetry [21]. Kinesin family motor proteins are involved in microtubule-dependent transport of a variety of cargoes and are required for the assembly of cilia in a process called intraflagellar transport [22]. Mouse knockouts of the kinesin gene Kif3B failed to assemble cilia and developed LR patterning defects [21]. Video microscopy revealed cilia in the ventral node in wild-type embryos beat with a clockwise vortical motion [23] to generate a right-to-left flow of extraembryonic fluid, which was termed ‘nodal flow’ [21]. Importantly, this asymmetric flow occurred just prior to the onset of asymmetric Nodal-Lefty-Pitx2 signalling. Mouse Kif3B knockouts lacked node cilia and nodal flow and developed defects in asymmetric Lefty expression and heart laterality. Studies using inversus viscerum (iv) mice with mutations in the left–right dynein (lrd) gene, which encodes the dynein axonemal heavy chain protein Dnah11 that is required for cilia motility, provided an additional link between paralysed node cilia and randomization of LR asymmetry [24–26]. More recently, an unbiased large-scale forward genetic screen identified mutations in 23 cilia-associated genes in laterality mutants [27], corroborating the critical role for cilia in LR patterning. The notion that directional fluid flows control LR patterning was demonstrated by the remarkable observation that Nodal-Lefty-Pitx2 asymmetry could be manipulated ex utero by applying artificial flow to mouse embryos in culture [28]. LR asymmetry in wild-type embryos was reversed with left-to-right artificial flows across the node, and laterality defects in iv mutants were rescued by applying normal right-to-left flows. These results support a model in which cilia-driven nodal flow is an essential cue that directs LR asymmetric expression of the Nodal-Lefty-Pitx2 cascade (figure 1). Although cilia are not required for LR asymmetry in all vertebrates (which is discussed in detail below), the finding that several vertebrate embryos have transient epithelial structures analogous to the mouse node that harbour motile cilia and asymmetric flows upstream of Nodal-Lefty-Pitx2 signalling suggests a conserved function for motile cilia in LR patterning [29]. These ciliated structures include (i) Kupffer's vesicle (KV), an ellipsoid organ in teleost fishes medaka [23] and zebrafish [30,31], (ii) gastrocoel roof plate, a flat triangular to diamond shaped epithelium in frog [32], and (iii) posterior notochordal plate, a convex ridge in rabbit [23]. Owing to the differences in names and locations, these analogous structures will be collectively referred to as the ciliated ‘left–right organizer’ (LRO).

Figure 1.

Functions of cilia in establishing vertebrate LR asymmetry. This schematic represents the transient LRO described in several vertebrate embryos that is comprised of epithelial cells that have either motile (yellow) or non-motile sensory cilia (green). The motile cilia beat with a vortical pattern and are posteriorly tilted to generate strong leftward fluid flows and slower rightward return flows. Leftward flows are essential for activating an asymmetric Ca²⁺ flux and the Nodal-Lefty-Pitx2 signalling cascade that controls LR asymmetry of developing organs. Three hypotheses have been proposed to explain how flows trigger asymmetric signals: (a) the ‘morphogen gradient model’ predicts that flows create a left-to-right concentration gradient of secreted signalling molecule(s) that bind receptors on the left side; (b) ‘nodal vesicular parcels’ (NVPs) that contain signalling molecules have been proposed to be transported to the left by flows; and (c) the ‘two-cilia model’ suggests that flows bend sensory cilia containing the ion channel Pkd2 to activate Ca²⁺ uptake on the left side. Recent results have challenged the idea of Ca²⁺-mediated mechanosensory functions for cilia, which is represented by (?). D = dorsal; V = ventral; L = left; R = right.

What do cilia-generated flows look like in LROs? In contrast with multiciliated cells that move mucus through airways in a single direction, monociliated cells in LROs create asymmetric flows in relatively small pools of recirculating extraembryonic fluid. Several innovative approaches have been developed to visualize and quantify these flows. Real-time tracking of fluorescent microbeads applied to the pit-like structure of the mouse node in cultured embryos revealed flows move leftward near the bottom of the ciliated pit and then return to the right near the top of the membrane-covered LRO [23]. Particle image velocimetry (PIV) analysis of flowing microbeads has been effectively used to measure flows in wild-type and mutant mouse embryos in culture and shed light on the relationship between flow strength and LR patterning (discussed more below) [33]. Analyses of microbead movements across the frog-gastrocoel roof plate using a semi-automated methodology called gradient time trails indicated microbeads that originate on the right margin are quickly moved to the left margin at specific developmental stages [32]. Understanding the dynamics of flows inside the fish LRO (KV) has been more challenging due to its more sphere-like shape. Anticlockwise rotational flow patterns have been described in KV by tracking fluorescent microbeads using manual [34] or automated [35] approaches or by tracking laser-induced cellular debris [36]. Observations of flow dynamics indicated strong right-to-left flow in anterior region of the LRO and slower return right to left flow at posterior pole [31,37,38]. More recently, automated analyses using customized ImageJ plugins [39] to track naturally occurring particles or a customized Matlab code [40] to track microbeads have generated ‘heat maps’ that show local flow intensities and confirm strong right-to-left flows in the anterior region. Thus, while imaging flows in living vertebrate embryos has revealed different dynamics in different species, the overarching theme of creating strong right-to-left flows is consistent among LROs (figure 1).

(b). Sensing asymmetric flows

Although solid evidence supports the role for cilia-driven directional fluid flows in establishing LR asymmetry in multiple vertebrates, how these flows are detected and translated into molecular asymmetries (e.g. Nodal-Lefty-Pitx2 expression) remains enigmatic. Several mechanisms have been proposed to explain how flows generate asymmetric signalling (figure 1). The ‘morphogen gradient model’ (figure 1a) proposes that directional flow sets up the left side of the LRO to receive a greater concentration of a secreted molecule(s) that triggers downstream asymmetric signalling. Indeed, fluorescent proteins introduced into the mouse node formed a left-to-right gradient [23], but endogenous secreted molecules/morphogens that direct Nodal-Lefty-Pitx2 expression have not been identified. Another model involves membrane-covered ‘nodal vesicular parcels’ (NVPs) (figure 1b) that contain signalling molecules that get swept to the left side of the LRO by cilia-driven flows [41]. Time-lapse imaging of cultured mouse embryos was used to track particles (NVPs) released into the node and then fragmented at the left wall. It was found that sonic hedgehog and retinoic acid were ensheathed in the NVPs and then released into the nodal flow in a fibroblast growth factor (FGF)-signalling-dependent manner. This is an attractive model that connects asymmetric flow with asymmetric delivery of signalling molecules, but a lack of additional studies beyond the initial observations of NVPs has left the regulatory details unclear.

As an alternative to the idea that asymmetric flows physically move signalling molecules, it was proposed that flow acts as a mechanical force to bend mechanosensory cilia on the left side of the embryo to activate signalling [42,43]. This ‘two-cilia model’ (figure 1c) posits that in addition to motile cilia that generate flow, there is a second population of non-motile sensory cilia that detect this flow. The initial evidence backing this model came from the discovery that mutations in the calcium ion (Ca²⁺) permeable channel protein Pkd2—which localizes to sensory cilia—disrupted LR patterning [44]. Subsequently, fluorescent Ca²⁺ indicator dyes and genetically encoded fluorescent Ca²⁺ reporters identified transient Ca²⁺ signalling on the left side of the LRO in mouse [42] and zebrafish [45] embryos just before the onset of asymmetric Nodal-Lefty-Pitx2 signalling. Loss of cilia-driven asymmetric flow [42] or Pkd2 ion channel expression [46], led to symmetric Ca²⁺ activity around the LRO and LR patterning defects, suggesting asymmetric Ca²⁺ flux is a key response downstream of flow. Based on observations that bending mechanosensory cilia on kidney cells allows Ca²⁺ to flow into the cell [47], it was hypothesized that stretch-sensitive Pkd2 channels in sensory cilia would open in response the mechanical stimulus of asymmetric flow on the left side of the LRO to establish left-sided Ca²⁺ flux. Indeed, Pkd2 protein localizes to LRO cilia in mouse [42], frog [32] and fish [48]. In the mouse LRO, Pkd2 was found in both motile cilia in the pit-like structure of the node and immotile cilia that are largely found in the surrounding crown cells [42,46]. Importantly, a mutant version of the Pkd2 protein that retained channel activity but could not localize to cilia failed to rescue asymmetric gene expression in Pkd2 mutant embryos [46]. Furthermore, when Pkd2 expression was reintroduced specifically in the crown cells of Pkd2 knockout mice, it rescued asymmetry of Nodal-Lefty-Pitx2 expression [46]. These results indicated the presence of Pkd2 in the cilium of cells surrounding flow-generating cells is important for LR signalling (figure 1c). Live imaging of fluorescent Ca²⁺ reporters in the zebrafish LRO identified Ca²⁺ flashes in cilia, called intraciliary calcium oscillations (ICOs), that preceded cytoplasmic Ca²⁺ signals in a Pkd2-dependent fashion [49]. This provided additional evidence that Pkd2 channels were opening in cilia to allow Ca²⁺ to enter the cell body and initiate Ca²⁺ flux in LRO cells.

Although the two-cilia model has garnered much support, recent studies have challenged the idea that cilia function as mechnosensors via Ca²⁺ signalling. First, work in cultured human cells indicated increased Ca²⁺ in primary cilia did not have a significant impact on cytoplasmic Ca²⁺ levels and did not stimulate Ca²⁺ release [50]. More recently, high-speed imaging of Ca²⁺ dynamics in transgenic mice expressing a Ca²⁺ indicator that localizes specifically to cilia showed that externally applied flows (at physiological levels or higher) failed to elicit Ca²⁺ response in primary cilia in diverse tissues [51]. Studies of the node in these transgenic mice revealed that physiological flow velocities barely deflected nodal crown cilia and did not evoke Ca²⁺ influx inside these cilia. It was suggested that previously observed Ca²⁺ signals within the cilium arise from Ca²⁺ diffusion from the cell body or entry due to damage to the cilia caused by external shear stress. Taken together these observations suggest that if cilia are mechanosensory, the mechanism does not involve Ca²⁺ entry through the cilium. Genetically encoded calcium sensors do have limitations [52], but these findings ask for a re-evaluation of the role of Pkd channels in LR patterning and a search for alternative ions and/or receptors that may be involved in sensing LRO flows.

Whether flows move morphogens or provide a mechanical stimulus remains an open and intriguing question. The mouse LRO contains between 200 and 300 motile cilia that produce strong leftward flows, and it was predicted that most of these cilia, if not all, were required to work in concert to establish robust flows and LR asymmetry. When movements of these cilia were retarded using a viscous material (methyl cellulose) in cultured mouse embryos, it was observed that weak and transient local flows were sufficient to initiate LR asymmetric gene expression [33]. This finding was confirmed in mouse cilia motility mutants Rfx3^−/− or Dpcd^−/− in which complete loss of flow disrupted LR asymmetry, but weak flows created by as few as two motile cilia were sufficient to generate normal LR signals [33]. Analyses and mathematical simulations of flow dynamics in the zebrafish LRO also predict only a subset of beating cilia (30 out of normal number of 50–60 motile cilia) are necessary to generate LR asymmetry [39], which correlates with an observed minimum size threshold for zebrafish LRO function [53]. Thus, looking forward, it will be important to determine whether weak localized flows can effectively transport morphogens or NVPs across the LRO, or activate mechanosensory mechanisms to regulate downstream asymmetric signalling events.

3. Cilia position matters

(a). Polarization of cilia

As hinted at above, the size and cellular architecture of LROs are quite different in different vertebrate species (see reviews: [54–56]). Despite these differences, one would predict that the arrangement and positioning of cilia in the LRO must be tightly regulated in order for these cilia to work in a coordinated fashion to generate effective fluid flows and LR signals. A long-standing question has been how clockwise vortical motion of LRO cilia generates a dominant unidirectional flow. Theoretical simulations of flow dynamics first suggested that motile monocilia are tilted posteriorly in the LRO [57]. Indeed, studies in mouse [23,58], rabbit [23], frog [32] and zebrafish [31,59] have identified posterior tilting of LRO cilia. High-speed imaging of cilia in the mouse LRO showed that the trajectory of cilia tips is shifted towards the posterior when compared with cilia roots [23]. This tilt allows a rotating cilium to make a leftward swing away from the cell surface and a rightward swing towards the cell surface. As the cell surface retards the movement of fluid by shear resistance, the leftward swing results in a higher degree of flow than the rightward swing, resulting in a net leftward flow [23,60].

This leads to the question of how cilia become tilted in LRO cells. Studies in mouse [23], rabbit [61] and frog [32] indicate cilia initially protrude from the centre of the apical surface of each cell, but then become posteriorly localized at stages that coincide with the onset of leftward flows. Such positioning of LRO cilia appears to be mediated by movement of the cilium base, or basal body, to the posterior pole of the cell [23]. Genetic studies indicate planar cell polarity (PCP) proteins are involved in the posterior migration of basal bodies in LRO cells. PCP refers to polarization of a field of cells in the plane of an epithelium first described in Drosophila [62]. Compound loss-of-function of the PCP genes Disheveled (triple knockout of Dvl1, Dvl2, Dvl3) or Van gogh like (double knockout of Vangl1 and Vangl2) in mouse embryos disrupted posterior positioning of LRO basal bodies and cilia [63,64]. This failure to tilt the cilia resulted in aberrant flows and LR defects. Protein localization studies revealed that the PCP proteins Vangl1 and Prickle2 accumulated on the anterior side of mouse LRO cells, whereas Dvl was found on the posterior [63,65]. These asymmetric localizations indicate LRO cells are polarized along the anteroposterior axis, but the mechanisms setting up this polarization are not understood [66,67]. Loss of Vangl2 in frog [65] and zebrafish [59] also disrupted LRO cilia posterior polarization, fluid flows and LR patterning. Collectively, these results indicate PCP-mediated posterior polarization of cilia is a conserved and essential mechanism for generating unidirectional flows and LR asymmetry.

(b). Positioning of ciliated cells

In addition to polarization and posterior tilt of LRO cilia, the arrangement of cells in the LRO plays important roles in generating fluid flows and LR signals. Emerging evidence points to distinct subpopulations of cells in LROs that must be properly positioned to produce and/or respond to flows. In mouse and frog LROs, cells with small apical surfaces and motile cilia are flanked on the left and right sides by much larger cells that express distinct molecular markers and show a mixed population of motile and immotile cilia [32,42,46]. Disruption of this arrangement in mouse Noto^−/− [68] or Zic3^−/− [69,70] mutants—such that the small cells are mixed with the larger cells—altered LR patterning. Further, Notch signalling regulated by the glycosylation factor Galnt11 determines the relative distribution of motile and non-motile ciliated cells in the frog LRO [71]. In zebrafish, the cellular architecture of KV is asymmetric along the anteroposterior and dorsoventral axes. KV cells in the dorsoanterior region develop an elongated columnar morphology with small apical surfaces—which allows tight packing of these cells—whereas ventroposterior cells adopt a cuboidal morphology with large apical surfaces [37,38]. This process, referred to as ‘KV remodelling’ [72], places more motile cilia in the dorsoanterior region to drive strong leftward flow across the anterior pole as described above. As observed in mouse and frog embryos, disrupting the arrangement of ciliated cells in the zebrafish LRO leads to flow and LR patterning defects [38,72,73].

Recent studies have identified regulators of the actomyosin cytoskeleton and components of the extracellular matrix (ECM) as mediators of LRO cell positioning. The Rho kinase protein Rock2, which operates downstream of Rho GTPase signals to modify the actomyosin cytoskeleton, was identified in a copy number variant screen in patients with heterotaxy syndrome [74]. Rock2 proteins are expressed in frog and zebrafish LROs, and loss-of-function studies confirmed their involvement in LR patterning [38,74]. In zebrafish, depletion of the Rock protein Rock2b or chemical inhibition of its target non-muscle myosin II did not affect cilia motility, but prevented regional cell shape changes during KV remodelling that position more cilia in the dorsoanterior region of the LRO [38,72]. This eliminated leftward flow and altered LR signalling. Regulators of actomyosin activity are also involved in establishing the cellular architecture of the mouse LRO. Loss of the Rho family GTPase Rac1 or the FERM domain protein Lulu/Epb4.1l5 disrupted the movements of ciliated columnar cells with small apical surfaces that must displace larger endoderm cells to reach the ventral surface of the embryo [75]. Mouse embryos with mutations in fibronectin and/or integrin have revealed roles for ECM in establishing normal node morphology and geometry [76,77]. Analyses of endogenous as well as ectopically induced KV in zebrafish embryos have also identified important roles for ECM proteins in positioning ciliated cells [73]. Interfering with laminin or fibronectin, which accumulate at the axial–paraxial boundary adjacent to densely packed dorsalanterior cells, inhibited KV remodelling that is critical for generating leftward flows [73]. Taken together, these observations suggest precise organization of cells generates an architectural asymmetry within the LRO that is critical for its function and regulation of LR asymmetry.

(c). Laterality cues for cilia?

While it is clear that tight regulation of the ciliated cells in LROs is important for establishing LR signalling, it is less clear whether cilia-generated flows provide the initial break of symmetry. First, cilia are not universally required to establish LR asymmetry in all vertebrates. In the chicken embryo, Hensen's node is site of the first asymmetric gene expression, but this structure is thought to lack motile cilia. Chicken Talpid3 mutants with cilia defects develop normal LR asymmetry [78–80], indicating a cilia-independent mechanism is used for breaking symmetry. Indeed, cell-tracking experiments revealed that asymmetric cell movements around Hensen's node set up asymmetric signalling in the chicken embryo [81]. Intriguingly, the node in the pig embryo also lacks cilia [81] and, unlike the mouse ventral node that is exposed to extraembryonic fluid, pig node cells are covered by a layer of cells referred to as subchordal mesoderm [82]. The node of the cow embryo is also enclosed in subchordal mesoderm [82], suggesting multiple vertebrates use cilia-independent mechanisms to establish LR asymmetry. Second, several events that impact LR asymmetry described in the frog embryo occur prior to the appearance and function of cilia in the LRO [83,84]. These include signalling between ectoderm and migrating mesoderm cells mediated by the heparan sulfate proteoglycan Syndecan 2 during gastrulation stages [85,86] and ion gradients established by the H⁺ /K⁺-ATPase [87] and vacuolar-type H⁺-ATPase [88] proton pumps that could drive asymmetric localization of determinants, such as serotonin [89], at early cleavage stages. It has been proposed [90] that intrinsic chirality of the cytoskeleton is used to asymmetrically transport molecules (e.g. ion pumps) [91] or mediate the LR distribution of differentially imprinted chromatids [92] to break symmetry in the early embryo. It is possible that early developmental events generate LR positional information that can be used to guide ciliated cells as they develop into a functional LRO or to completely bypass the requirement for cilia [93,94].

Several connections have been made between molecules involved in early LR events and the development of the LRO: Syndecan 2 [95] and the vacuolar-type H⁺-ATPase [88,96,97] are involved in the development of normal LRO morphogenesis and cilia formation in zebrafish, and H⁺/K⁺-ATPase activity [98] and serotonin [99] are needed for Wnt signalling pathways that control LRO development and ciliogenesis in frog embryos. One outstanding question is whether precursor cells receive laterality cues prior to forming the LRO. As discussed above, the location of different types of cells within the LRO is critical for their functions; however, it is not known whether these subpopulations differentiate and organize at the time of LRO morphogenesis or bring with them fates that were determined at earlier stages of development. To begin to address this question, we are taking advantage of the zebrafish embryo to investigate the behaviours of precursor cells that give rise to the LRO. Unlike other vertebrates, fate-mapping studies [100,101] have identified all of the precursors—called dorsal forerunner cells (DFCs)—that form the zebrafish KV/LRO (figure 2). Transgenic embryos that express GFP in the DFCs using a sox17 promoter [102] have been widely used to track the development of these cells. During gastrulation stages, DFCs migrate together toward the vegetal pole of the embryo and then compact into a tight a cluster. Connections with overlying cells provide multiple foci of apical membrane and junction proteins [103]. DFCs then undergo rearrangements that bring the foci into close proximity to form a rosette-like structure that gives rise to the nascent lumen where cilia begin to elongate (figure 2a). Formation of rosette-like structures has also been described during early morphogenesis of the mouse node/LRO [75]. However, it is not known whether precursor cells randomly incorporate into these rosettes or take up pre-determined positions.

Figure 2.

Tracking precursor cells of the zebrafish left–right organizer Kupffer's vesicle. (a) Schematic of the development of dorsal forerunner cells (DFCs) into Kupffer's vesicle (KV) in the zebrafish embryo. A cluster of DFCs at the 90% epiboly stage or 9 h post-fertilization (hpf) rearranges to form a rosette-like structure with a central point of apical membrane proteins at the bud stage (10 hpf). These cells epithelialize, form cilia and begin to expand the fluid-filled lumen by the two-somite stage (11 hpf). Red and green cells represent mosaic labelling of DFCs using a loxP-Cre approach (see electronic supplementary material methods). (b) Mosaic labelling of DFCs in Tg(ubi:Zebrabow);Tg(sox17:Cre^ERT2) double transgenic embryos imaged at the 80–90% epiboly stages. Long treatments with 4-hydroxytamoxifen (4-OHT) from dome stage to 90% epiboly stage (approx. 5 h) resulted in widespread Cre activity in DFCs that converted most or all of the cells from expressing default red fluorescent protein (RFP) to yellow fluorescent protein (YFP). Short 4-OHT treatments from the dome stage to 60% epiboly stage (approx. 2.5 h) labelled only a few DFCs with YFP expression. Embryos not treated with 4-OHT had no YFP⁺ cells. Both vegetal and side views are shown for each embryo. Scale bars = 20 µm. (c) Time-lapse imaging was used to analyse individual YFP⁺ DFCs during development. Three-dimensional reconstructions of YFP⁺ DFCs that did not divide were tracked for 60 min. starting approximately 9.5 hpf. In this representative analysis, labelled cells #1 and #2 maintained their relative positions during the DFC to KV transition. Scale bars = 50 µm. Dotted lines are approximate boundaries of DFC clusters.

To investigate the behaviour of cells during this process, we developed a Cre-loxP–based mosaic cell-labelling method (see electronic supplementary material methods) to monitor individual DFCs in living zebrafish embryos (figure 2b). Time-lapse microscopy was used to track labelled DFCs during the process of clustering, rosette formation and lumen expansion. Interestingly, this analysis revealed little or no mixing of cells during this process: a cell located in a specific quadrant of the DFC cluster retained that position during formation of KV (electronic supplementary material, Movie 1). For example, a cell that originated in the left-anterior quadrant of the precursor pool became a ciliated cell in the left-anterior quadrant of the LRO (figure 2c). These observations suggest precursor cells are not stochastically arranged in the LRO as it forms, but rather these cells acquire positional information—and potentially specific cell fates—at earlier stages of development. It will be interesting to test this hypothesis in future studies and investigate factors that may provide laterality cues for cilia during development.

4. Conclusion

Understanding how, when and where LR asymmetry is established in vertebrate embryos remains a fascinating problem that holds significant biomedical relevance. The discovery of cilia motility defects in patients with LR malformations has provided insight into how the asymmetric body plan is set up during embryogenesis, and work using animal models has identified roles for cilia in generating left-sided Nodal-Lefty-Pitx2 asymmetry. But, exactly how cilia generate LR asymmetry is not yet clear. One important unanswered question is how weak fluid flows produced by only a few motile cilia are detected and translated into molecular signals. It remains to be determined whether mechanosensitive cilia, morphogen gradients or vesicular particles—which are not mutually exclusive mechanisms—are involved in sensing flows. New studies are also needed to further clarify how functions of cilia are influenced by the cellular topography of the LRO. In addition, it will be important to determine whether precursors of these ciliated cells are asymmetrically patterned and/or specified into distinct subpopulations prior to the appearance of the LRO. Answers to these questions will be critical for advancing our understanding of how cilia generate LR asymmetry in the vertebrate body plan.

Ethics

All zebrafish experiments were conducted with approval by SUNY Upstate Medical University's Institutional Animal Care and Use Committee.

Data accessibility

Methods and data supporting figure 2 have been uploaded as part of the electronic supplementary material.

Authors' contributions

Experiments were designed by J.D.A. and A.D. A.D. performed these experiments. A.D and J.D.A. analysed the data and wrote the manuscript.

Competing interests

We have no competing interests.

Funding

This work was supported by grants from the National Heart, Lung and Blood Institute (R01HL095690) and National Institute of General Medical Science (R01GM117598).