Planar cell polarity signaling in the development of left–right asymmetry

Abstract

The planar cell polarity (PCP) signaling pathway, principally understood from work in Drosophila, is now known to contribute to development in a broad swath of the animal kingdom, and its impairment leads to developmental malformations and diseases affecting humans. The ‘core’ mechanism underlying PCP signaling polarizes sheets of cells, aligning them in a head-to-tail fashion within the sheet. Cells use the resulting directional information to guide a wide variety of processes. One such process is lateralization, the determination of left–right asymmetry that guides the asymmetric morphology and placement of internal organs. Recent evidence extends the idea that PCP signaling underlies the earliest steps in lateralization and that PCP is invoked again during asymmetric morphogenesis of organs including the heart and gut.

Introduction

Planar cell polarity (PCP) signaling controls the polarization of cells within the plane of an epithelium or within mesenchymal tissue, orienting asymmetric cellular structures, cell divisions, and cell migration. Much of our mechanistic understanding of PCP signaling derives from work using Drosophila as a model system. In flies, PCP signaling controls the orientation of hairs on the adult cuticle, chirality and orientation of ommatidia in the eye, orientation of cell divisions, and related processes in other tissues. While much effort has been focused on mechanistic studies in flies, medically important developmental defects and physiological processes in vertebrates are also under control of PCP signaling, motivating considerable interest in studying PCP in vertebrate model systems. Defects in the core PCP mechanism result in a range of developmental anomalies and diseases including open neural tube defects (reviewed in Copp and Greene [1]), deafness (reviewed in Montcouquiol and Kelley [2]), situs inversus and heterotaxy and associated heart defects (reviewed in Santos and Reiter [3]) and were implicated in polycystic kidney diseases (reviewed in Simons and Walz [4]), although that notion has more recently been challenged [5]. PCP polarizes skin and hair [6] and the ependyma [7]. PCP is also believed to participate in both early and late stages of cancer progression (reviewed in VanderVorst et al [8]) and during wound healing [9]. The PCP component Prickle, although apparently not other PCP pathway components, is mutated in an epilepsy-ataxia syndrome [10]. Mutations in global PCP components have recently been associated with a human disorder of neuronal migration and proliferation [11] and congenital kidney defects [12].

Work in Drosophila has identified and partially characterized a ‘core’ module that responds to directional cues to generate and amplify asymmetric subcellular localization of PCP signaling molecules and to coordinate polarization between neighboring cells. The core module responds to one or more ‘global’ modules that provide directional information to orient core module polarization with respect to tissue axes, and it instructs tissue-specific downstream effectors that direct morphogenetic responses. Some vertebrate tissues use a core mechanism very similar to that in Drosophila to generate polarization, whereas other systems, sometimes orchestrating morphologically diverse functions, may implement PCP differently [13].

Here, I describe the core PCP signaling mechanism and its role in the establishment of left–right asymmetry and in heterotaxy-associated developmental abnormalities.

The core module in Drosophila

The Drosophila core PCP module has two essential, mechanistically interlocking, functions. Molecular asymmetry of epithelial cells orthogonal to their apical basal axis is generated by segregating distinct molecular complexes to opposite sides of the cell. Simultaneously, it aligns the polarity of each cell with that of its neighbors through interaction of these complexes between adjacent cells.

The proximal core proteins include the 4-pass protein Van Gogh (Vang; aka Strabismus), the peripheral membrane Prickle Espinas Testin (PET/Lim) domain protein Prickle, and the seven-pass atypical cadherin Flamingo (Fmi; aka Starry night), and the distal core proteins include the serpentine protein Frizzled (Fz), the peripheral membrane proteins Dishevelled and Diego, and Fmi (Figure 1A) (reviewed in Butler and Wallingford [14]). Intercellular Fmi–Fmi bridges facilitate interaction between proximal and distal complexes on adjacent cells to form intercellular complexes. Intercellular complexes assemble into an intrinsically asymmetric structure, such that heterotypic proximal— distal interactions are favored over either homotypic interaction, and proximal and distal complexes segregate to opposite sides of the cell owing to their proclivity to exclude each other from adjacent areas of the cell surface [15]. Modifications including phosphorylation, ubiquitinylation, neddylation, and intracellular trafficking appear to play roles in this process. The net result is that the mechanism behaves as a bistable amplifier; beginning from randomly distributed proteins and complexes and given molecular noise, individual cells establish a proximal–distal axis and locally align their axes with those of their neighbors. In the absence of an overlaid directional input, swirling patterns of short- but not long-range local continuity will form. However, given a global directional bias, well-aligned patterns oriented to the tissue axes form over a large field (Figure 1B and C) [16].

Figure 1.

Schematic of core PCP protein interactions and illustration of their function in the Drosophila wing. (a) Interactions between oppositely oriented proximal (Vang, Pk, and Fmi)–distal (Fz, Dsh, Dgo, Fmi) complexes induce their segregation to opposite sides of each cell (right). The resulting distribution polarizes each cell and links the polarity of neighboring cells, creating local polarity correlation. (b) A wing with weakened global polarity input and a local polarity disruption produces swirling patterns displaying local polarity but significant regions of globally misoriented polarity (ft¹/ft^Grv; dpp-GAL4/UAS-fz). Hairs on the adult wing emerge from the side of the cell where the distal complex localizes and points in the direction defined by a vector pointing from the cell center toward the distal complex. (c) An adult wing with intact global and core PCP signaling produces a largely parallel array of hairs. Dgo, Diego; Dsh, Dishevelled; Fmi, Flamingo; Fz, Frizzled; PCP, planar cell polarity; Pk, Prickle; Vang, Van Gogh.

Although the specific interactions among core PCP proteins that result in the intrinsic asymmetry of intercellular complexes and in the mutual exclusion of oppositely oriented complexes are incompletely understood, the emergent properties of the system are well established. Current understanding of the molecular and cell biology of this mechanism has been recently and thoughtfully reviewed [14].

The bistability inherent in the core module allows for any systematic input bias to direct coordinated polarization in a specified direction. One amply characterized global signal, directed by the atypical cadherins Ft and Ds and the Golgi-resident protein four-jointed (Fj) (reviewed in Matis and Axelrod [17]), and several other less well-defined sources of global directional information, including the proposed activity of Wnt proteins and possibly other unidentified signals, appear to act in partially overlapping, tissue-dependent ways [18,19]. Their modes of action, and current controversies, are beyond the scope of this review. In brief, the Ft/Ds/Fj system is proposed to read expression gradients to direct subcellular trafficking of core module components [20,21], and a separate, apparently, Drosophila-specific mode of action links more directly to the core module [22,23]. Similarly, diffusion gradients of Wnts are proposed to provide instructive signals that act on Fz [18].

Conservation of PCP signaling in vertebrates

Although the nature of experimental systems limits analyses, it is now well established that the core PCP module is functionally conserved in at least some vertebrate (and other chordate) systems. Of defining importance, the asymmetric interdependent subcellular localization of core proteins, as well as planar-polarized responses to core component activity, have been detailed in a variety of settings. In vertebrates, apparent complexity is increased by multiple paralogous homologs of single core PCP genes in flies, resulting in examples of both functional redundancy and diversification of function within a single polarization process. Nonetheless, existing data are consistent with the inference that the emergent property of bistable polarization and coordination of polarity is retained in these settings (reviewed in Vladar et al [13] and Butler and Wallingford [14]). In other instances, evolutionary divergence appears to have co-opted and modified the core PCP mechanism to operate differently in performing diverse functions. These variations will not be further discussed here.

Little data exist concerning global signals and connectivity in vertebrates, but there is likely to be a diversity of systems at this level as well, particularly considering the much greater scale of polarizing tissues. Homologs of Ft, Ds, and Fj exist, and phenotypes suggest similar roles at least for FT4 and DCHS1, but their potential contributions are not well understood. In some instances, mechanical forces have been proposed to orient core PCP signaling to tissue axes. A persistent theme is the contribution of Wnt proteins in vertebrate PCP signaling. However, compelling evidence that they provide instructive rather than permissive signals has been elusive, and the participation of non-Fz receptors adds a layer of complexity (reviewed in Butler and Wallingford [14]).

Initiation of left–right asymmetry

Left–right asymmetry, either externally manifested or confined to internal organs, is a noteworthy feature among bilaterians. In Drosophila, this asymmetry is observed in the embryonic hindgut and again in the adult hindgut and male terminalia, each of which undergoes dextral looping during its morphogenesis [24]. In humans, left–right asymmetry is observed in both the positioning and morphology of many organs. Asymmetry of the heart is first observed early in development and allows for the differentiation and connectivity of the systemic and pulmonary circulation. The left curvature of the gut initiates its stereotypical coiling and compaction within the abdominal cavity. PCP plays a critical role in the earliest known lateralization event in bilaterians and then plays a recurrent role in later generation of organ asymmetry. Failure to correctly determine left–right laterality disrupts situs solitus (the normal asymmetric arrangement of internal organs), resulting in either situs inversus, the complete reversal of lateralization, or heterotaxy, the partial randomization of organ laterality which is invariably associated with cardiac malformations and often other organ defects [25].

The earliest transcriptional evidence of bilaterian left–right asymmetry is the conserved left-sided expression of Nodal, a transforming growth factor (TGFb) family member, its inhibitor, Lefty, and the transcription factor Pitx2. Multiple feedback loops sustain and propagate this expression on the left side of the lateral plate mesoderm (reviewed in Blum et al [26]) (Figure 2A). In a range of bilaterian embryos, the signal that initiates this left–right transcriptional asymmetry originates in the left–right organizer (LRO), a transient structure located at the posterior end of the notochord at early somite stages. Although of varying morphology, the LRO typically is an epithelium composed of cells with motile and nonmotile cilia extending into a fluid-filled compartment. As first observed in the mouse, motile cilia in the LRO (aka node) produce a leftward fluid flow [27] that determines the direction of lateralization [28]. It was subsequently found that central motile cilia are surrounded by nonmotile cilia that produce calcium fluxes coincident with the flow [29]. These observations, taken together, led to the model that central motile cilia produce a flow-dependent signal that is sensed by peripheral nonmotile cilia, initiating a left side–specific transcriptional response. Whether flow is detected as a mechanical signal or by transporting a chemical signal remains unresolved [30]. Although LRO architectures vary among species, motile ciliated structures are broadly conserved and appear to function similarly [26,31].

Figure 2.

Cell chirality and generation of left-right asymmetry in the LRO. (a) Posterior localization on the dome-shaped apical surface (inset) determines the posterior tilt of the cilia and thus the leftward fluid flow within the LRO. Leftward flow stimulates peripheral cells via their nonmotile cilia, producing a left side–specific calcium flux that initiates left side–specific nodal signaling. (b) Proximal (green) and distal (red) core PCP complexes distribute to the anterior and posterior sides of motile ciliated cells in the LRO of many bilaterians. Their localization is read out to determine the posterior localization of the ciliary basal bodies (blue) and thus the motile cilia on the apical surface. Gradients of Wnts and Wnt inhibitors (Sfrp’s) appear to instruct the orientation of PCP polarization along the anterior–posterior axis in the mouse, whereas tissue stress is thought to instruct direction in the frog. LRO, left–right organizer; PCP, planar cell polarity.

While ciliary motility and leftward fluid flow are required, it remained unclear how cilia might create a directional flow. Motivated by theoretical arguments, it was found that motile cilia tilt posteriorly as they rotate owing to their posterior positioning on dome-shaped cellular apices [32]. The resulting eccentric beating motion of the cilium was shown by mechanical simulation and in silico modeling to produce the predicted leftward flow [32,33]. To achieve posterior localization and tilt of cilia, the central LRO cells must distinguish their anterior from posterior sides, and this naturally led to examination of a potential role of the PCP pathway in generating this asymmetry. Several reports subsequently demonstrated that central LRO cells show asymmetric localization of core PCP proteins and that mutation of the core PCP system disrupts the posterior cilium localization, disrupting leftward flow and the subsequent lateralization of the Nodal/Pitx2 pathway in the mouse, frog, and zebrafish [34–36] (Figure 2B). These observations place core PCP signaling at the very earliest steps of left–right lateralization, providing vectorial information within the developing LRO. More recently, several models for global directional cues that orient this PCP have been proposed. In the mouse LRO, gradients of the noncanonical Wnts Wnt5a and Wnt5b and several Sfrp’s, inhibitors of Wnt signaling, resulting from localized perinodal expression, may provide the global directional information needed to orient the core PCP system [37*]. In contrast, in the Xenopus LRO, anisotropic strain resulting from gastrulation movement appears to instruct the orientation of PCP complexes along the anterior-posterior (A–P) axis [38*].

Surprisingly, some LROs, specifically Hensen’s node in the chick (and likely other avian species and reptiles) and pig, are nonciliated and therefore do not produce fluid flow but nonetheless initiate the Nodal/Pitx2 pathway [26]. Instead, in Hensen’s node, leftward-biased cell migration is thought to underlie left–right specification [39]. PCP is implicated in regulating this migration [40], suggesting that some bilaterians have lost the leftward flow mechanism yet retained PCP as a foundational aspect of lateralization. Curiously, this echoes recently observed cell flows during PCP-dependent polarization of murine hair follicles [41*].

Organ looping and heterotaxy

Once the Nodal/Pitx2 pathway is initiated, lateralization of internal organs responds to this cue. For example, in the chick and mouse, differential changes in cellular morphogenesis between left and right sides of the dorsal mesentery directed by Pitx2 initiate chiral gut looping [42], and left–right differences in cell migration determine rotation and leftward lateral movement of the zebrafish developing heart cone in response to the Nodal homolog Southpaw [43]. It is of great interest to define the effector pathways that differentially respond to these signals in these tissues. While the Nodal signal undoubtedly biases lateralization, the observation that heart looping occurs when Southpaw signaling is blocked, and even occurs in explanted Southpaw-mutant heart tubes, supports the existence of a tissue-intrinsic, nodal-independent mechanism contributing to chiral heart looping [44]. This tissue-intrinsic mechanism may normally be biased by the Nodal pathway to determine directionality.

What might be the molecular basis for this tissue-intrinsic looping ability? To address this in vertebrate models, lessons from Drosophila have proven instructive. In Drosophila, the heart tube is symmetric, but chiral rotation of the adult hindgut depends on core PCP activity within the rotating gut [24]. Interestingly, the organizer of Drosophila adult hindgut rotation requires the unconventional myosin Myo31DF, encoding Drosophila MyoID, to determine directionality, and Dachsous, with which it physically interacts, to transmit this directional signal to the core PCP system. The target on which PCP acts in the adult hindgut is unknown, but importantly, before looping, cells adopt a biased orientation of their cell junctions and long axes with respect to the gut axis (Figure 3). A potential causal relationship between these events is hinted at by the observation that in Myo31DF mutants, cells adopt the opposite orientation before looping in the incorrect direction. Although not directly addressed, these observations suggest the possibility that core PCP signaling organizes cellular chirality that presages and may contribute to the looping event.

Figure 3.

Model for PCP-directed cell chirality in developing gut and heart tubes. Evidence suggests that cells in the developing gut or heart tube are or may be polarized by PCP signaling, defining their anterior–posterior axes. The hypothetical localization of posterior PCP complexes is shown in red. With the apical–basal polarity axis also defined, intrinsic cell chirality can determine left–right directionality. Cell chirality is apparent in the biased, off-axis orientation of cell shapes and in the asymmetric localization of E-cadherin (Drosophila gut) or N-cadherin and phospho-myosin II (chick heart) localization (blue). Looping of the heart and gut may be driven, at least in part, by either relaxation of the elongated cell shapes or biased junctional rearrangements. An extrinsic left-sided signal from Nodal can extinguish cell chirality, providing further asymmetry to the morphogenetic process. If PCP protein asymmetry is present, as expected, there is no specific expectation about whether Nodal signaling might alter or extinguish it. PCP, planar cell polarity.

Although the importance of cell chirality in adult hindgut looping has not been addressed, it has been more fully studied in the earlier and distinct looping of the embryonic hindgut. Before embryonic hindgut rotation, cell shapes show a left–right–biased chirality as defined by the position of the centrosome to the cell centroid and the orientation of cell junctions relative to the long axis of the gut [45]. Furthermore, E-cadherin is asymmetrically enriched and is required for the establishment or maintenance of cell chirality (Figure 3). The left–right bias of each of these features is reversed in Myo31DF mutants. Strikingly, mosaic analysis shows that this is a cell-intrinsic property [46]. Experimental evidence indicates that the force driving hindgut looping is generated intrinsically, but unexpectedly, a minimal amount of junctional rearrangement is observed [45,47*]]. These observations, coupled with simulations, suggest a model in which the relaxation of pre-existing chiral cell shapes leads to twisting of the embryonic hindgut tube [47].

Although it is tempting to connect the dots and speculate that core PCP signaling organizes and coordinates the chiral cell shapes that then drive looping, there remain several gaps in this model. First, asymmetric subcellular localization of core PCP components, the signature of core PCP activity, has not been reported in the embryonic or adult gut. Second, in the adult hindgut, where core PCP activity is required, its relationship to cell chirality has not been tested. Third, the core PCP system, while required in adult hindgut looping [24], is not needed for embryonic hindgut looping [45]. Thus, some other signal, in conjuction with Myo31DF, must determine the direction of cell chirality. Potential contributions of Dachsous and Fat in embryonic hindgut looping have not been reported, so one purely speculative possibility is that the embryonic hindgut has features analogous to the adult hindgut organizer, in which Dachsous interacts with Myo31DF to produce a directional signal. In contradistinction to the adult hindgut organizer, that signal would be directly translated into cell chirality rather than acting nonautonomously through the core PCP system. Therefore, for now, the evident relationship between PCP, cell chirality, and gut looping in Drosophila remains ill defined.

Themes derived from Drosophila, including appearances of cell chirality and core PCP participation re-emerge in studies of organ lateralization in vertebrates. In the zebrafish and mouse gut, left–right differential expression of multiple PCP components is observed, and activity is required for lateralization in the dorsal mesentery to direct gut rotation, suggesting that side-specific PCP functions intrinsically to differentially modify cellular morphogenesis to control gut looping [48]. Similarly, mouse heart looping depends on the PCP components Dvl2, Vangl2, and Wnt5a [49]. Furthermore, core PCP signaling was shown to regulate actomyosin activity to control cell neighbor exchange, and regionally controlled cell neighbor exchange in the distal ventricle and outflow tract are required for zebrafish heart looping [50*]. In Xenopus, Vangl2 interacts genetically with MyoID to affect lateralization, including heart looping, suggesting deep conservation at this level, although at which step this interaction occurs was not defined [51*]. While heart development is complex and differs in these models, these observations suggest PCP signaling and cell chirality may contribute to intrinsic mechanisms of heart lateralization.

In addition to conserved roles for PCP and MyoID, the picture has been extended by recent observations in cultured chick heart cells showing that intrinsic cell chirality is modulated by Nodal signaling [52*]. As assessed by Golgi positioning and myosin II enrichment in the heart tube before looping, cells on the right side show a greater chirality relative to the left side. Dramatically, this cell chirality and the difference between left and right was replicated when myocardial cells were disaggregated and plated in matrigel cultures. Cells from the right side showed a strong bias to rotate in a clockwise direction, whereas cells from the left side approximated random rotation. Activation of Nodal signaling, or activation of aPKC signaling, suppressed and even reversed the clockwise rotation in vitro, and atypical Protein Kinase C (aPKC) activation suppressed their chirality in vivo, suggesting a mechanism by which Nodal can bias the intrinsic asymmetry that likely drives looping. As in the Drosophila model, the relationship between cell chirality and PCP signaling remains to be explored.

Disease connections

Given the association of PCP with heterotaxy in model systems, it is important to ask whether mutations in the PCP signaling pathway are associated with laterality defects. To date, such mutations have not been associated with heterotaxy or with heterotaxy-related congenital heart defects, the most common morbidity related to heterotaxy [25]. This is not, however, necessarily indicative of a lack of association. In light of this finding, it is instructive to look at another well-studied spectrum of PCP-related congenital malformations. It has long been known that open neural tube defects in animal models result from disruption of core PCP signaling. PCP signaling instructs neuroepithelial and mesenchymal cells in defining their A-P axes and regulates convergent extension, a process of neighbor exchange driving the lateral intercalation of cells that underlies neural tube closure. This process and the role of PCP in regulating it have been reviewed elsewhere [14,53]. In a perhaps important parallel, PCP signaling is thought to control asymmetric actomyosin contractions that drive convergent extension [54]. It was therefore surprising that PCP mutations were not readily associated with neural tube defects. However, with the introduction of more powerful screening technologies, it has emerged that PCP mutations are indeed significantly associated with these defects, only in a polygenic fashion [55,56]. Perhaps, combinations of hypomorphic variants are compatible with life and can result in malformation, whereas stronger hypomorphs and amorphs lead to inviability. It is therefore tempting to speculate that polygenic PCP mutations will yet be found responsible for a subset of heterotaxy and congenital heart malformations as well.