Much of what is known about planar cell polarity (PCP) has been revealed through work in Drosophila melanogaster. The two most rigorously studied examples are PCP's role in polarizing wing hairs (trichomes) to distally pointing parallel arrays, and in eliciting rotation of ommatidia toward the equator of the eye disk (Strutt et al., 1997). PCP also polarizes epithelia in the adult thorax (Lawrence et al., 2002), imaginal disks (Strigini and Cohen, 2000), and the sensory organ precursor cell (Gho and Schweisguth, 1998). In addition to regulating patterning events, PCP is also required for proper morphogenetic movements of epithelia during germ band elongation (Lecuit et al., 2002) and dorsal closure (Kaltschmidt et al., 2002).
PCP also polarizes epithelia in vertebrates, but its functions in vertebrate patterning have largely been revealed in just the past 5 years. In vertebrates, PCP positions outer hair cell stereociliary bundles in the mouse inner ear, orients cell divisions during gastrulation (Gong et al., 2004), and is required for morphogenetic movements during neural tube closure (Heisenberg et al., 2000; Wallingford and Harland, 2002), convergent extension (cell intercalation leading to embryonic axis extension; Heisenberg et al., 2000; Wallingford et al., 2000), and convergence of bilateral endoderm and heart precursors to the midline (Matsui et al., 2005).
Defects in PCP are readily visible by means of abnormal positioning of cellular components that are normally polarized. For example, PCP mutant clones in the Drosophila wing lead to hairs that point in all directions rather than lining up in ordered arrays (Fig. 1A,B). Similarly, mouse PCP mutants display misorientation defects in stereociliary bundles of the inner ear (Fig. 1C,D). Most likely, PCP polarizes most, if not all, epithelia. However, without the benefit of a readily visible marker to demonstrate a cell's orientation, it is difficult to ascertain the end result. Therefore, there are likely many more PCP-dependent cellular events yet to be discovered.
CORE PCP PROTEINS
The PCP pathway is composed of several “core” proteins that, by mechanisms that are not yet understood, induce intracellular cytoskeletal rearrangements in response to an extracellular polarity cue (Fig. 2; Wong and Adler, 1993). In vertebrates and in Drosophila, the polarity cue is mediated by Frizzled (Fz), a seven-pass transmembrane receptor (Gubb and Garcia-Bellido, 1982; Vinson et al., 1989; Park et al., 1994). Although Fz can function as a receptor for wingless (Wg), PCP signaling in Drosophila is independent of Wg (Lawrence et al., 2002). The unknown diffusible Fz ligand has been dubbed “Factor X,” although whether or not Factor X actually exists is under debate (see below). Paradoxically in vertebrates, PCP is downstream (although perhaps indirectly) of noncanonical Wnts, Wg orthologs that do not signal by means of β-catenin (Moon et al., 1993; Heisenberg et al., 2000; Tada and Smith, 2000; Dabdoub et al., 2003).
Upon activation, Fz signals to the core PCP effectors: Dishevelled (Dsh), a cytoplasmic protein that binds Fz; a negative regulator of the pathway, the LIM-domain protein Prickle (Pk); the atypical cadherin Flamingo/Starry Night (Fmi/Stan); the transmembrane protein Strabismus/Van Gogh (Stbm/Vang); and the ankyrin repeat protein Diego (Dgo; Taylor et al., 1998; Wolff and Rubin, 1998; Chae et al., 1999; Gubb et al., 1999; Usui et al., 1999; Feiguin et al., 2001). Homologs of all of these genes, except diego, have also been identified and shown to mediate vertebrate PCP (Sumanas et al., 2000; Tada and Smith, 2000; Wallingford et al., 2000; Park and Moon, 2002; Carreira-Barbosa et al., 2003; Curtin et al., 2003; Takeuchi et al., 2003; Veeman et al., 2003). In Drosophila, core PCP proteins function in a complex network to regulate one another's cellular localization. In the wing and eye, core PCP proteins are initially distributed uniformly along the apical cortex of the cell, but over time they become localized to distal (Fz and Dsh), proximal (Pk and Stbm), or both membranes (Fmi, Dgo; Usui et al., 1999; Axelrod, 2001; Feiguin et al., 2001; Shimada et al., 2001; Strutt, 2001; Tree et al., 2002b; Bastock et al., 2003). Core PCP proteins physically interact with one another, and localization of core proteins are dependent on one another (Strutt, 2001; Tree et al., 2002b; Bastock et al., 2003; Wu et al., 2004), suggesting they may form a multiprotein complex that positions one another asymmetrically in the cell.
How does Fz signaling confer directionality? Fz has cell autonomous as well as cell nonautonomous effects that are temporally separable (Strutt and Strutt, 2002). The implication is that Fz not only functions intracellularly to process directional information, but that it also helps produce a polarity cue. Taken together with the data presented above, two models have been proposed. One hypothesis is that a source of Factor X activates Fz, and in turn Fz signaling generates a short-range signal (Factor X or another molecule) that subsequently propagates PCP between cells (Adler et al., 1997, 2000). The second model is that there is no diffusible Factor X. Instead, Fz activity triggers a “domino effect,” whereby asymmetric localization of core proteins in one cell leads to the asymmetric distribution of core proteins on adjacent cells (Adler et al., 1997; Tree et al., 2002a; Ma et al., 2003). The second model is dependent on the supposition that asymmetric localization is required for propagation of PCP, a notion that has not yet been tested (Lawrence et al., 2004). Finally, it is possible that both models are correct and that different epithelia rely on each mechanism to different extents (see below).
Fj, Ds, AND Ft FUNCTION UPSTREAM OF Fz
In Drosophila, Four-jointed (Fj) is a type II transmembrane protein that, along with the large atypical cadherins Dachsous (Ds) and Fat (Ft), lie genetically upstream of Fz and the signal that promotes the overall directionality of PCP in an epithelia (Zeidler et al., 1999a, b, 2000; Rawls et al., 2002; Yang et al., 2002). Although there are putative homologs of these three proteins in vertebrates, roles for them in vertebrate PCP have not been reported (Dunne et al., 1995; Matsuyoshi and Imamura, 1997; Ashery-Padan et al., 1999). In Drosophila, the interplay between Fj, Ds, and Ft is just beginning to be understood. fj acts upstream of ds, and in addition, there is positive feedback loop between a fj/ds/ft pathway and fj (Yang et al., 2002). The finding that Fj is expressed in the Golgi and that Golgi-tethered Fj is capable of driving PCP (Strutt et al., 2004) supports a model whereby Fj regulates the activity of Ds and possibly Ft through posttranslational modifications. Furthermore, Ds antagonizes Ft (Adler et al., 1998), and both Ft and Ds localize to membranes and mediate heterophilic cell interactions (Strutt and Strutt, 2002; Ma et al., 2003; Matakatsu and Blair, 2004). These findings raise the possibility that Ds functions as a ligand and Ft as a receptor on adjacent cells.
Work has also shown that Ft physically and genetically interacts with the transcriptional corepressor, Atrophin (Atro/Grunge), and both ft and atro repress fj expression (Fanto et al., 2003). Based on these observations it was proposed that, upon activation of Ds by Fj, the repressor activity of Atro is inhibited, thereby resulting in the activation of Factor X or Fz. Although an intriguing model, it is dependent on propagation of PCP being dependent on transcription, something that remains to be tested.
Events upstream of Fz may diverge in different types of epithelia. Ds and Fj are expressed in opposing gradients in the wing, eye, abdomen and imaginal discs (Zeidler et al., 2000; Rawls et al., 2002; Yang et al., 2002; Strutt et al., 2004), leading to the hypothesis that graded expression of Ds and Fj are required for graded activity of Factor X or Fz. This idea is supported by the finding that reversal of Ds and Fj gradients promote a reversal in ommatidia polarization in the eye (Simon, 2004). However, it was also found that, when Ds expression is rendered uniform in the wing, PCP is unaffected (Matakatsu and Blair, 2004). These findings suggest that events upstream of Fz may be tailored to function differently in distinct cellular environments. For example in the eye field, ommatidia are interspersed with other cell types that apparently do not respond to and cannot propagate Factor X. In this situation, it would stand to reason that a gradient of Factor X, set up by Fj and Ds expression gradients, may help confer directionality. By contrast, all wing cells respond to PCP; therefore, directionality information may be transferred directly from cell-to-cell, rendering Fj and Ds gradients unnecessary.
DOWNSTREAM OF CORE PCP PROTEINS
What causes polarized cellular changes downstream of core PCP components remains largely unknown. The best-studied molecular pathways that function genetically downstream of fz and dsh are EGF (epidermal growth factor), Notch, Rho GTPase, and Rac GTPase. The EGF pathway is critical for ommatidia rotation, and the EGF inhibitor Roulette, as well as the MAP kinase Nemo, and the Ras effector Canoe/AF6 (Choi and Benzer, 1994; Gaengel and Mlodzik, 2003), regulate this process. The Notch receptor and its ligand Delta regulate asymmetric photoreceptor cell fate decisions in ommatidia (the R3/R4 cell fate decision; Gho and Schweisguth, 1998; Cooper and Bray, 1999; Fanto and Mlodzik, 1999; Tomlinson and Struhl, 1999), and/or R3/R4-specific behavior (Strutt et al., 2002). Numb, a negative regulator of Notch signaling, is also required for orientation of asymmetric sense organ precursor cell divisions (Gho and Schweisguth, 1998). Rac1 GTPase signals through JNK (jun N-terminal kinase) to regulate ommatidia rotation and wing hair polarization in a pathway parallel to Rho GTPase and Drok (Drosophila Rho-associated kinase; Strutt et al., 1997; Fanto et al., 2000; Winter et al., 2001).
Thus far, only Rho and Rac GTPases have identified conserved roles as PCP effectors in vertebrates. In Xenopus and zebrafish, the Rac1-JNK pathway also functions in parallel to a Rho-Rho Kinase 2 pathway to regulate the polarized morphogenetic event, convergent extension (Habas et al., 2001, 2003; Marlow et al., 2002; Tahinci and Symes, 2003). Because PCP can produce different cellular outputs, one might expect downstream PCP effectors to vary between cell types; therefore, additional effectors may yet be revealed.
Downstream of Drosophila RhoA, Drok provides a connection to the proposed major output for PCP, cytoskeleton rearrangement. Drok regulates phosphorylation of nonmuscle myosin regulatory light chain, resulting in a conformational change in the nonmuscle myosin II (Winter et al., 2001). In germ band elongation, myosin II activity is required to remodel cell junctions and for cell intercalation, two polarized cellular changes that drive this morphogenetic event (Bertet et al., 2004). Thus, there is evidence for a direct but incomplete pathway between Fz, Rho, Drok, nonmuscle myosin, and cytoskeletal rearrangement leading to polarization-dependent cell behaviors.
A CONVERSATION WITH THE EXPERTS
There are many unsolved mysteries in the field of PCP research. Some of these are discussed below in an interview with two PCP scientists, Matthew Kelley, Ph.D., Acting Chief, Section on Developmental Neuroscience, NIDCD/NIH, Bethesda, Maryland, and Helen McNeill, Ph.D., Senior Investigator, Samuel Lunenfeld Research Institute, Toronto, Canada (Fig. 3). An extended version of this discussion can be viewed at http://www.interscience.wiley.com/jpages/1058-8388/suppmat
Dev. Dyn.:What is your favorite model to explain establishment and propagation of PCP?
H.M.: The model for PCP propagation by means of asymmetric localization of PCP components was based on the observation that these components align on one side of wing epithelial cells and that disruption of these components in a clone of cells clearly perturbs the localization on adjacent cells. It is an elegantly simple model that works very well for the wing. However, PCP in the eye is much more complicated. The asymmetric accumulation of PCP proteins is only visible on ommatidial precluster cells, and there is a sea of unpolarized cells separating the polarized cells. Therefore, I think that there is still a need for a “Factor X,” at least in some tissues. I favor a model whereby the mysterious Factor X regulates Fz activity, which leads to the spatially restricted accumulation of PCP proteins (Fz, Dsh, Pk, Stbm, and so on). This then spatially remodels the cytoskeleton and cell surface to generate visible PCP.
M.K.: In the inner ear, I think that the data are not consistent with the need for a Factor X, or not yet. Hair cells do not directly contact one another, suggesting that signaling from cell-to-cell would be difficult. But, we do not know about the expression and localization of PCP genes in the intervening supporting cells, largely because we do not have sufficiently good antibodies yet. The supporting cells are not obviously polarized, but this does not necessarily mean anything, and in fact, there are several supporting cell types that have clear polarization suggesting that the signal could be passed from cell to cell.
H.M.: Yes, a signal could be passed from cell to cell without being visible. However, the model was based on the visible accumulation seen in the wing, and that is why it is so persuasive in the wing. Such evidence is lacking from the eye, and its more complicated structure suggests a need for additional mechanisms.
Dev. Dyn.:Do you think the basic mechanisms of PCP varies between tissues or model organisms?
H.M.: I definitely think that there is a high degree of conservation in PCP mechanisms. It is overwhelming the number of genes that have been shown to function in Drosophila PCP that also function in vertebrate convergent extension and neural tube closure. I suspect that there is a core PCP pathway that functions in all tissues and model organisms and that divergence will occur mainly at the tissue specific level.
It is clear from work in Drosophila that there are several genes that function in PCP in the eye and not in the wing; some examples include the JAK/STAT pathway, the Iroquois transcription factors, and the Notch pathway. It is interesting that several of the genes that appear to affect PCP in the eye (but not the wing) also regulate vertebrate PCP. In addition, of course, there will doubtlessly be vertebrate specific elaborations of the PCP pathway—but the core and logic of the pathways will probably be the same in both flies and man.
Dev. Dyn.:Why do you think the identity of Factor X in Drosophila has been so elusive? Do you think noncanonical Wnts in vertebrates are functionally orthologous to Factor X?
H.M.: It is possible that Factor X has been hard to find because it has other functions earlier in development that mask its later role in PCP. It is common for important signaling molecules to be used again and again during development, so there is certainly precedence for this problem.
Factor X is believed to give positional information to cells, directing their planar polarity. Because ubiquitous expression of Wnt11 can rescue some of the PCP-like defects of the Wnt11 mutant silberblick (slb), this Wnt, at least, cannot be Factor X. Also, because there is so much conservation, and because the fly data suggests that Wnts are not the PCP ligand, I suspect that the same is true for vertebrates.
M.K.: We do have some data that are tantalizing in terms of Wnts and Factor X. Within the cochlea, Wnt-7a is asymmetrically expressed at one side of the sensory epithelium, and either gain or loss of Wnt signaling results in similar changes in hair cell PCP. But, the Wnt signaling pathway is so complex that there are several other possible explanations. Wnt-7a might only be permissive rather than instructive or it has been shown that changes in the level of Wnt signaling can indirectly affect the PCP pathway as a result of recruiting important signaling molecules, such as Dishevelled, away from the PCP pathway. We are doing some experiments to try to reverse the gradient of Wnt-7a in the cochlea, but absent any experimental data, my feeling at this point, given the fly data, is that Wnt/Wg is probably not Factor X.
H.M.: I think the case of Wnts in the fly eye has some relevance here—it is pretty clear that Wg is not the PCP activator of Fz in the eye. But Wg controls the expression of Dachsous and Four-jointed and Mirror; genes that do act upstream of Fz in PCP in the eye. So, although the expression of Wg is suggestive, and it does regulate PCP indirectly, it does not seem to be acting as the PCP ligand, at least in the fly eye.
Dev. Dyn.:What are some important remaining questions in the field that need to be addressed?
M.K.: In vertebrates, there is a huge amount of work to be done. We have only identified a handful of the important molecular players and we have very little understanding of the molecular interactions. Moreover, we do not understand how the asymmetric signal is converted into a change in the cells at a biological level. For instance, in the inner ear, the ultimate effects of defects in PCP genes is the misplacement of the stereociliary bundle, but we do not understand at a cell biological level how the cell places the bundle to begin with, or how that machinery is disrupted in PCP mutants. In invertebrates, I think that a greater number of the players have been identified, but there are still many questions regarding how the signal is conveyed throughout the plane of an epithelium.
H.M.: Yes, more of the players have been identified in flies, but there are still enormous gaps. And for the most part the gaps are the same in flies and vertebrates. How is Fz activity regulated? How does the PCP signal result in the restructuring of the cytoskeleton and cell surface of cells? How does it regulate cell movements and their final morphology? Like Matt, I think that we have very little knowledge of the cell biology of PCP, and some of the most exciting research in the next few years will be filling in that gap.
M.K.: This is a really exciting time for PCP. The community has assembled enough tools that I think we have reached a point where the rate of knowledge acquisition will really start to increase. But there is so much more to learn which makes for an intellectually challenging environment.