Planar cell polarity in the Drosophila eye is directed by graded Four-jointed and Dachsous expression

Epithelial planar cell polarity (PCP) is present when the cells of a tissue are polarized along a common axis lying in the plane of the epithelium(Eaton, 1997). In many tissues, the organized planar polarization of cells is essential for differentiated function. For example, in the vertebrate auditory system, the sensory hair cells of the inner ear are polarized in a uniform direction so that their stereocilia bundles can respond to sound(Yoshida and Liberman, 1999). In other cases, planar polarization is used to control cell fate decisions either by orienting asymmetric cell divisions(Adler and Taylor, 2001) or by regulating signaling events between adjacent cells(Cooper and Bray, 1999; Fanto and Mlodzik, 1999; Tomlinson and Struhl, 1999). The lack of obvious local directional cues within the plane of an epithelium has led to the proposal that the axis of polarization is controlled by long-range signaling molecules that diffuse to form gradients across the tissue (Lawrence, 1966). Identifying these diffusible signals and understanding how their action orients planar polarization are central challenges in the study of PCP.

Much of our knowledge of PCP comes from studies of the Drosophilawing, where PCP is apparent in the orderly arrangement of the actin hairs that grow from the distal edge of each cell(Adler, 2002; Shulman et al., 1998). These studies have shown that the site of hair growth is designated by the presence of high levels of Frizzled (Fz) signaling activity along the distal edge of each cell. This polarization state results from a competitive process whereby adjacent cells compare Fz signaling activity levels across their proximodistal junctions (Adler et al., 1997). An essential feature of this competition is the presence of feedback loops that amplify any differences in Fz activity between two adjacent cells(Tree et al., 2002). As a result of these feedback loops, the side of the cell-cell junction with a higher initial level of Fz activity develops stronger Fz signalling, while the opposing side reduces its Fz activity(Strutt, 2001). In this model,the uniform orientation of wing PCP arises from signals that consistently bias the Fz competition so that the proximal side of each cell-cell junction always emerges with elevated Fz levels. The identity of these directional signals is currently unknown.

Another example of PCP in Drosophila occurs in the compound eye,where PCP is evident in the orientation of the dorsal and ventral ommatidia,which are mirror images of each other (Fig. 1A-D) (Wolff and Ready,1993). To form this pattern, developing ommatidia must sense the positions of the equator and/or the nearest pole, and polarize in response. Evidence that ommatidial polarization is mechanistically similar to PCP in the wing comes from observations indicating that a core group of proteins essential for PCP in the wing is also required for properly oriented ommatidial polarity (Adler,2002; Shulman et al.,1998).

Ommatidial orientation is tightly coupled to a photoreceptor cell fate decision within each ommatidium. Ommatidial differentiation begins in the eye imaginal disc and is marked by the progression from posterior to anterior of the morphogenetic furrow (Fig. 1E). Anterior to the furrow, the position of the equator has been determined but ommatidia have not begun to form(Papayannopoulos et al.,1998). Posterior to the furrow, ommatidial assembly begins with the differentiation of photoreceptor R8, followed by photoreceptors R2 and R5,and then the precursors of photoreceptors R3 and R4. At this stage the developing dorsal and ventral clusters are indistinguishable. Mirror image polarity arises when the adjacent R3/R4 precursor cells of each precluster adopt distinct fates, such that the one located closest to the equator becomes R3, while the polar cell becomes R4. These cell fate decisions are determined by a PCP competition in which the equatorial cell emerges with higher Fz activity (Tomlinson and Struhl,1999; Zheng et al.,1995). Once the R3 and R4 cells have been specified, the dorsal and ventral clusters rotate 90° in opposite directions, and add the remaining photoreceptors to establish the final mirror-image polarity pattern.

Although considerable insight has been gained into the Fz PCP signaling competition, the signals that directionally bias this competition have been more elusive. The key role of Fz in the establishment of PCP originally suggested that the axes of planar polarization in the wing and eye might be specified by Wnt gradients across the tissue. The resulting gently graded activation of Fz across the tissue could be refined by Fz/PCP competition at each cell-cell boundary to generate the final pattern of polarized Fz signaling. However, searches for Wnts with the expected expression patterns,high in the proximal region of the wing and in the equatorial region of the eye, have not identified appropriate candidates.

During a previous study, we proposed an alternate model for orienting planar polarization in the eye (Yang et al., 2002). In this model, the role of the long-range diffusible PCP signals is to generate opposing transcriptional gradients of the cadherin Dachsous and the Golgi-associated protein Four-jointed(Fig. 1F). The resulting gradients of Ds and Fj protein were proposed to regulate the function of the ubiquitously expressed cadherin Fat (Ft), resulting in the equatorial R3/R4 precursor cell of each ommatidium having a higher level of Ft function than its polar neighbor. This difference in Ft activity then biases the Fz/PCP competition between the R3/R4 precursor cells to ensure that the equatorial cell emerges with the higher Fz signaling state. Similarly, a study of Fj, Ds and Ft function in the abdomen has suggested that gradients of Fj and Ds expression may direct planar polarization in that tissue as well(Casal et al., 2002).

This model, which was derived from an analysis of the consequences of removing Fj, Ds or Ft function from one of the R3/R4 precursor cells, makes three crucial predictions with regard to the effects of altering the expression patterns of Fj, Ds and Ft. First, because the directional cues for orienting PCP in the eye are proposed to be present in the Fj and Ds expression gradients, the organized pattern of polarization should be lost when both gradients are replaced by ubiquitous expression. Second, reversing the graded pattern of Fj or Ds expression should lead to corresponding reversals of ommatidial polarity. Finally, ectopic expression of Ft in a graded pattern might be able to specify an altered direction of polarization. In this report, I test and confirm each of these predictions. I further show that the remarkable fidelity of PCP in the eye results from the combined action of the Fj and Ds gradients acting through Ft.

The following alleles were used: fj^d1 (Zeidler et al., 1999), fj^N7 (Buckles et al.,2001), ds^38K(Clark et al., 1995) and ds^UA071 (Adler et al.,1998). The ds^38K and ds^UA071 mutations are strong loss-of-function alleles that show little if any detectable staining with anti-Ds antisera(Yang et al., 2002). The fj^d1 allele is a null allele, whereas the fj^N7 allele is a strong hypomorph. Transheterozygous allelic combinations were used in all crosses to avoid problems that arose from unrelated mutations present on the various mutant chromosomes. The Gal4 drivers used were TubP-Gal4 (Lee and Luo,2001), Omb3-Gal4 and Vg1-Gal4(Tang and Sun, 2002). The Vg1-Gal4 driver insertion is also a fj mutant allele.

The UAS-Fj construct used was previously described(Zeidler et al., 1999). For Ds, a genomic DNA fragment extending from 35 bp before the translational initiation codon to the end of exon 2 was linked in frame to a genomic fragment containing the remaining coding exons and extending 1933 bp beyond the termination codon. These joined fragments were then placed into the UAS-containing P-element vector pUASp(Rorth, 1998) as a NotI fragment. The P-element insert used resides on chromosome 3. For Ft, the genomic region extending from 60 bp before the start codon and extending 2734 bp beyond the termination codon was placed in pUASp as a NotI fragment. The P-element insert used resides on chromosome 3. This transgene efficiently rescues the lethality of ft flies, as indicated by the recovery of greater than 80% of the expected number of ft^G-rv/ft^l(2)fd; TubP-Gal4/UAS-Ftadults. Complete sequences of the inserted Ft and Ds fragments are available on request.

Sectioning was performed as previously described, except that the osmium steps were omitted (Tomlinson and Ready,1987). Staining with anti-Ft antibodies was performed as previously described (Yang et al.,2002). Wings were examined after mounting in Euperol.

A central feature of our model is the proposal that the orientation of the opposing Fj and Ds expression gradients determines the direction of planar polarization in the eye. Thus, the replacement of both gradients with ubiquitous expression should lead to the absence of PCP directional cues, and result in a randomized pattern of dorsal and ventral ommatidia. To test this prediction, a ubiquitously expressed TubP-Gal4 driver was used to drive expression from UAS-Fj- and UAS-Ds-containing P-elements in fj and ds mutant backgrounds (Brand et al., 1994). For Ds, this required the generation of a P-element construct in which the ds genomic coding region, with the large second intron removed, was placed under the control of UAS transcriptional regulatory elements. For Fj, a previously characterized UAS-Fj transgene was used (Zeidler et al.,1999).

In order to evaluate the roles of the Fj and Ds gradients, the pattern of ommatidial polarity was examined for various combinations of loss-of-function,ubiquitous expression and wild-type expression of each gene. The animals tested and the resulting phenotypes are summarized in Table 1. Several of the results are crucial to evaluating the roles of the Fj and Ds expression gradients. First, the overall pattern of polarization is retained when either the Fj or the Ds gradient is replaced with ubiquitous expression(Fig. 2E,F). However, in each case, a low rate of polarity errors is observed(Table 1). This contrasts with wild-type eyes in which polarity mistakes are rarely, if ever, observed(Fig. 2A). These findings indicate that neither expression gradient is essential for directing PCP in the eye. However, the occurrence of polarity mistakes suggests that the strength or quality of information is reduced when one gradient is replaced by ubiquitous expression. By contrast, the replacement of both gradients with ubiquitous expression results in the complete loss of organized PCP, as indicated by a randomized pattern of polarity in which dorsal and ventral type ommatidia are mixed together without an obvious equator(Fig. 2G). Taken together,these results demonstrate that the Fj and Ds expression gradients provide the directional cues that specify the orientation of PCP in the eye, but that the directional information provided by the two gradients is partially redundant. Furthermore, these data indicate that both gradients are required to provide the robust directional cues needed to achieve the perfect fidelity of polarization observed in wild-type eyes.

The analysis of the effects of replacing the Fj and Ds expression gradients with ubiquitous expression also demonstrate an important distinction between the roles of Fj and Ds during PCP signaling in the eye. For Fj, the only role indicated is to provide directional information via its gradient of expression. This limited role is apparent in the observation that animals either lacking Fj function or having ubiquitous Fj expression display equivalent phenotypes. Either condition leads to a reduced fidelity in the presence of a wild-type Ds gradient (Fig. 2C,E), and to a loss of organized polarity when Ds is expressed ubiquitously (Fig. 2G,H). In contrast to Fj, Ds has an essential role during the establishment of PCP that can be separated from the role of its expression gradient in providing a directional cue. This additional role is evidenced by the lack of organized PCP seen in ds animals (Fig. 2B) (Rawls et al.,2002; Yang et al.,2002), as well as by the ability of ubiquitously expressed Ds to rescue the ds phenotype when graded Fj expression is present to provide the required directional cue (Fig. 2F versus 2G,H).

In order to provide further evidence that the orientation of the Fj and Ds expression gradients specified the axis of PCP in the eye, I tested whether reversing the expression gradients would lead to corresponding reversals of ommatidial polarity. For this purpose, the previously described Vg1-Gal4 and Omb3-Gal4 drivers were used to express Ds and Fj(Tang and Sun, 2002). The Vg1-Gal4 driver is an insertion in the fj gene and drives expression in a graded fashion that is high at the equator and low towards the poles. The Omb3-Gal4 driver, located in the optimotor blind gene, expresses Gal4 strongly at the extreme poles of the eye and rapidly declines towards the equator.

Ectopic expression of Fj using either the Vg1-Gal4 or the ubiquitous TubP-Gal4 drivers had little effect on the pattern of ommatidial polarity(Fig. 3A,B). In each case, only occasional mis-specified ommatidia were seen. These results were expected as neither driver reverses the wild-type Fj gradient. By contrast, expression of Fj under the control of the Omb3-Gal4 driver had profound effects in both wild-type and fj animals. When the resulting eyes were sectioned to reveal the area near the dorsal pole, large-scale reversals of ommatidial polarity were observed (Fig. 3C,D) (see also Strutt et al.,2004). Importantly, these reversals were located in the region where the expression of the Omb3-Gal4 driver declines rapidly. The contrast between the effects of Vg1-Gal4- and TubP-Gal4-driven Fj expression, and Omb3-Gal4-driven Fj expression, demonstrate that reversing the Fj expression gradient, but not merely overexpressing Fj, can override the normal PCP cues and specify a new orientation of polarity. Consequently, these data strongly support the proposal that graded expression of Fj is a source of PCP directional cues in wild-type eyes.

Similar conclusions were supported by experiments in which Ds was ectopically expressed. Although the expression of Ds using either the Omb3-Gal4 or TubP-Gal4 drivers had only minor effects on the pattern of ommatidial polarity (Fig. 3E,F), Vg1-Gal4-driven Ds expression resulted in the frequent appearance in the equator region of ommatidia having reversed polarity(Fig. 3G). This effect was further strengthened in ds animals, which lack the opposing endogenous Ds gradient (Fig. 3H). The resulting eyes displayed disorganized polarity consistent with the idea that coherent directional cues were absent due to the opposing information provided by the endogenous Fj and the ectopic Ds gradients. In order to better visualize the effects of Ds expression in an equator to polar gradient, Vg1-Gal4-driven Ds expression was examined in ds, fjanimals, which lack endogenous directional cues. The equatorial regions of these animals frequently displayed a clearly perceptible equator in which most of the nearby ommatidia were reversed in polarity and thus pointed towards rather than away from the equator (Fig. 3I). Away from the equator, the pattern of PCP in these animals become more randomized, perhaps due to a lack of sufficient Ds expression to provide its essential PCP function (data not shown). As with Fj, the ability of graded, but not ubiquitous, Ds expression to re-orient PCP strongly supports the role of the endogenous Ds gradient in orienting PCP in wild-type flies.

The results presented above demonstrate that the establishment of graded Fj and Ds expression represent an essential step in providing the directional information that orients PCP in the eye. In our earlier work, we proposed that these expression gradients then direct ommatidial polarity by producing a subtle equatorial gradient in Ft activity. As a result of this Ft gradient,the equatorial R3/R4 precursor cell within each ommatidium would have a slightly higher level of Ft function than the adjacent polar precursor cell. This consistent difference in Ft activity within each ommatidium was proposed to bias the Fz competition to ensure that the equatorial precursor cell always assumes the high Fz signaling state. An expectation of this model is that strongly graded ectopic expression of Fat, which is normally expressed evenly across the eye imaginal disc, might be capable of overriding the normal directional signals and specifying a new orientation of polarization.

In order to test this prediction, a P-element transgene was created that placed Ft expression under the control of UAS transcriptional regulatory elements. This transgene was used in combination with the TubP-Gal4, Vg1-Gal4 and Omb3-Gal4 drivers to provide different patterns of Ft expression. As Ft plays an essential role in the regulation of epithelial proliferation, the Vg1-Gal4 or OmbGal4 drivers were used to provide additional expression rather than replacing endogenous Ft function(Bryant et al., 1988; Mahoney et al., 1991). This resulted in overall gradients of expression. The altered gradients produced are shown in Fig. 4A-C. It is worth noting for these experiments that the Vg1-Gal4 driver is substantially stronger than the Omb3-Gal4 driver (data not shown).

TubP-Gal4-driven expression of Ft efficiently rescued the viability of animals otherwise lacking Ft function(ft^G-rv/ft^l(2)fd). Examination of the eyes of the rescued flies showed normal overall polarity with only occasional incorrectly polarized ommatidia (Fig. 4D). Similar results were also obtained when the TubP-Gal4 driver was used to express Ft in otherwise wild-type animals(Fig. 4E). Thus, overexpression of Ft is not sufficient to substantially alter the effectiveness or function of the normal PCP directional cues. Similarly, expression of Ft using the Vg1-Gal4 driver also had little effect on the pattern of PCP in the eye(Fig. 4F). This finding was expected as the Vg1-Gal4 driver would merely reinforce the equator-to-pole gradient of Ft activity predicted by the model.

Unlike TubP-Gal4- or Vg1-Gal4-driven expression, expression of Ft using the Omb3-Gal4 driver consistently led to polarity reversals(Fig. 4G). In each of ten eyes examined, a small group of ommatidia (from 2-5) displayed reversed polarity. These reversed ommatidia were found in the region of the eye where Omb3-Gal4-driven Fj had its effects, and which corresponds to the region where there is a decline of the Omb3-Gal4-driven gradient. These results suggested that providing a pattern of Ft that is higher at the poles provides a directional PCP cue that can overcome the cues provided by the endogenous Fj and Ds expression gradients. However, this effect is weak. In order to more clearly demonstrate the ability of a Ft expression gradient to provide a directional PCP cue, the effect of the Omb3-Gal4-driven Ft was examined in the absence of Fj function, which weakens the opposing endogenous directional information. A marked strengthening of the polarity reversal phenotype was observed under these conditions, with large groups of ommatidia near the poles co-ordinately reversing their polarity(Fig. 4H). These results confirm that the creation of a gradient of Ft function through ectopic expression is capable of providing a directional signal, and support the idea that the normal polarization pattern results from a high equator/low polar gradient of Ft activity.

Several lines of evidence have suggested that gradients of Fj and Ds expression might also orient planar polarization in the Drosophilawing. These include the expression of Fj and Ds in opposing patterns, as well as the severe polarization defects observed in ds animals(Fig. 5B,F)(Adler et al., 1998; Ma et al., 2003; Strutt and Strutt, 2002; Zeidler et al., 2000). In addition, fj and ds mutant clones often show non-autonomous effects on the polarity of nearby cells in a manner reminiscent to the effects seen near clones of cells mutant for genes whose products are known to participate in PCP signaling (Adler et al.,1998; Zeidler et al.,2000). Finally, clones of ft mutant cells in the wing display polarity defects consistent with the idea that the mutant cells are unable to directly sense global directional PCP cues(Ma et al., 2003).

In order to determine whether the Fj and Ds gradients have an important role in wing PCP, I examined the effects resulting from either replacing both gradients with TubP-Gal4-driven expression, or else replacing of the Ds gradient concomitant with the loss of fj function. In the eye,animals with either genotype have a randomized pattern of ommatidial polarity(Fig. 2G,H). Surprisingly,examination of these same animals revealed normal polarity throughout the wing blade with the exception of a small region along the anterior proximal margin of the wing where a small swirl is often present(Fig. 5C,D). These results demonstrate that the Ds and Fj expression gradients do not provide essential directional information in the wing.

The conclusion that the Ds and Fj expression gradients do not play essential roles in orienting PCP in the wing was also made in a recent report(Matakatsu and Blair, 2004). However, that report did not directly assay animals in which both gradients of expression were absent. Instead, strong ubiquitous Ds overexpression was used to in an attempt to overwhelm the endogenous Ds gradient in fjanimals. Using our UAS-Ds expression construct, such conditions conditions(w¹¹¹⁸; fj^N7/ds^UA071, fj^d1;TubP-Gal4/UAS-Ds) do not lead to a loss of organized PCP in the eye,although the PCP fidelity rate is considerably reduced (85%). As a result, the experiments presented here represent a more rigorous test of the role of the graded expression of Fj and Ds during the establishment of PCP in the wing.

The development of organized PCP requires cells to polarize in response to directional signals within the plane of the epithelium. The apparent absence of local cues has suggested that cells orient their polarity in response to long-range diffusible signaling molecules that form gradients across the tissue. We previously proposed that the role of the diffusible signals, such as Wingless produced at the poles of the eye disc(Wehrli and Tomlinson, 1998),is to drive graded transcription of Ds and Fj(Yang et al., 2002; Zeidler et al., 1999). In this model, the resulting Ds and Fj protein gradients then regulate the function of the cadherin Ft, resulting in a Ft activity gradient, which in turn controls the pattern of Fz competition within each ommatidium. At the time of that report, crucial tests of the model were precluded by our inability to alter the patterns of Ds and Ft expression. In this study, I have analyzed the effects of altering Fj, Ds and Ft expression in the eye, and have provided evidence supporting crucial features of the model. Most importantly, I have demonstrated that the Fj and Ds expression gradients provide redundant directional information that together orient PCP. Furthermore, the data shows that it is the combination of both gradients that provides the robust directional cues needed to support the perfect fidelity of polarization in wild-type eyes. In addition, I have shown that graded Ft expression can direct the pattern of ommatidial polarity, thus providing support for the role of Ft as a graded regulator of Fz signaling acting under the control of the Fj and Ds gradients.

In our proposed model, the consistent equatorial bias of Fz signaling results from more effective Ft action in each equatorial R3/4 precursor cell when compared with its adjacent polar counterpart. As this Ft difference results from the action of the Fj and Ds gradients, a key question is how these gradients could control the level of Ft function. Important insight into this issue has come from studies of the wing that suggest that Ft and Ds form a complex in which the localization of Ft on the surface of one cell is promoted by binding to Ds on the surface of the neighboring cell(Ma et al., 2003; Matakatsu and Blair, 2004; Strutt and Strutt, 2002). The dependence of Ft plasma membrane localization on Ds may account for the requirement for Ds function during planar polarization in the eye even when sufficient directional cues are provided by the Fj expression gradient.

The existence of Ds:Ft intercellular dimers suggests several mechanisms by which Ds might regulate Ft. One simple possibility is that Ds merely controls the accumulation of Ft on the surface of the neighboring cell. Thus, the relatively higher level of Ds in the polar R3/R4 precursor, which results from the polar gradient of Ds expression, would lead to the accumulation of more Ft on the bordering surface of the equatorial cell. This would result in an asymmetry in Ft protein levels precisely along the border between the precursor cells where Fz/PCP competition occurs. Although no such gradient has been observed, it would certainly be very subtle and perhaps undetectable. A second possibility is that Ds binding to Ft regulates Ft activity rather than localization. A third possibility is that Ds could participate with Ft in binding to the extracellular domain of a downstream target.

Fj appears to play a more limited role than Ds during planar polarization of the eye. Unlike Ds, which both contributes a directional signal through its graded expression and plays an essential role in the interpretation of directional cues, Fj appears only to participate in PCP establishment via the directional information provided by its graded expression. This more limited role can be seen in the observations that either the absence or the ubiquitous expression of Fj yields equivalent phenotypes, and does not grossly disrupt the pattern of polarization unless the Ds gradient has been replaced with ubiquitous expression. How might graded Fj fulfill this role? One possibility,proposed by Ma et al. (Ma et al.,2003) and Strutt et al.(Strutt et al., 2004), is that Fj may regulate the ability of Ft and Ds to productively interact with each other. Thus, the higher expression of Fj in the equatorial cell of each ommatidium leads to more Ft:Ds dimers being formed with Ft in the equatorial cell than in the opposite orientation. As Fj appears to function in the Golgi,this regulation may involve the direct modification of Ft or Ds(Strutt et al., 2004).

It is important to note that one aspect of the data reported here requires reconsideration of a feature of our previous model(Yang et al., 2002). In our previous work, we proposed that Fj acts upstream of Ds, perhaps by modifying the Ds activity gradient. This placement was based on genetic experiments showing that strong differences in Fj activity between R3/R4 precursor cells can direct ommatidial polarization only when Ds is present. The identification of an essential gradient-independent function for Ds clearly complicates the interpretation of these epistasis experiments. As a result, it is no longer possible to infer whether the information provided by the Fj expression gradient acts upstream of Ds to modify the information provided by the Ds gradient. An equally plausible possibility is that Fj regulates the function of the Ds:Ft complexes by modifying Ft rather than Ds function.

The work presented here was designed to test specific predictions of the model proposed in our earlier study. However, alternate roles for Ft function have also been proposed (Fanto et al.,2003; Rawls et al.,2002). In one model, Ft regulates the production of an unidentified long-range signal that is secreted at the equator and that directly controls eye polarity (Rawls et al., 2002). The existence of such an unidentified patterning signal, often called Factor X, has been invoked frequently to explain the`domineering nonautonomy' phenomenon seen in both the wing and the eye near clones of cells lacking function of PCP genes such as Fz (reviewed by Adler, 2002). In the alternate model, the role of Ft is to prevent production of this factor everywhere in the eye except at the equator where Ft activity is proposed to be inhibited by unspecified mechanisms, presumably involving Ds. An important distinction between the two models relates to the predicted effects of graded Ft expression. In our model, graded Ft activity provides the key PCP directional cues, and thus ectopic Ft expression gradients are predicted to have the potential to orient ommatidial polarity. In the alternate model, gradients of Ft activity do not provide directional cues. Instead, it is the lack of Ft activity in a sharp zone at the equator that leads to the production of the unidentified patterning factor. As a result, this second model predicts that subtle gradients of Ft expression should not orient polarity, especially in the polar regions of the eye where Ft activity is proposed to be uninhibited. Thus, the data presented in this report demonstrating the orienting ability of Ft expression gradients presents a challenge to this alternate model. In addition, the need for Factor X, whose putative existence has been a common feature of PCP models in both the wing and eye, has been challenged recently on both experimental and theoretical grounds(Ma et al., 2003) (J. Axelrod and C. Tomlin, personal communication). These reports have suggested that domineering nonautonomy results from the tendency of neighboring cells to align their polarization rather than the existence of an additional polarizing signal.

The key roles of Ft and the Fj and Ds expression gradients in the eye naturally raised the question of whether similar mechanisms are used to provide directional cues in other tissues, such as the wing. That such conservation might exist was suggested by the existence of gradients of Fj and Ds in the wing. Additionally, it has been demonstrated recently that ectopic gradients of Ft and Ds expression in the wing can produce re-orientation of polarity in the wing (Matakatsu and Blair,2004). Given the redundant nature of the directional cues provided by the Fj and Ds gradients in the eye, the most rigorous way to evaluate the roles of the Ds and Fj expression gradients in the wing was to examine the consequences of removing the directional information of both gradients simultaneously. When this was done, the resulting wings displayed almost completely normal polarity. Thus, the Ds and Fj expression gradients do not play a major role in orienting PCP in most of the wing blade. One possibility is that there are additional directional signals that act redundantly with the Ds and Fj gradients. Another possibility is that these gradients exist for reasons unrelated to PCP. For example, they may serve to regulate the function of Ft as a regulator of cellular proliferation(Bryant et al., 1988). Possible support for such a role comes from the observation that flies in which both graded Fj and Ds expression has been replaced with ubiquitous expression survive to adulthood at reduced frequencies, and often display defects in the size and shape of their legs, wings and eyes (data not shown).

The dispensability of the Fj and Ds gradients of expression during the polarization of the wing indicates that there must be currently unidentified directional cues directing wing PCP. Despite their mysterious nature, it is likely that their mode of action will involve the Ds:Ft complex. This inference can be drawn from the observation that animals lacking Ds function,or clones of cells lacking Ft or Ds activity, have substantial PCP defects in the wing. Importantly, clones of ft mutant cells in the wing appear not to read directional cues and instead align their polarity with that of their neighbors (Ma et al.,2003). Thus, whatever the nature of the unidentified signals, they appear not to function effectively in the absence of Ds and Ft. As neither Ft nor Ds is directly required for the Fz PCP signaling at cell-cell junctions,the dependence of these unidentified signals on Ds and Ft suggests that they may act by asymmetrically modifying the action of the Ds:Ft complexes at cell-cell junctions engaged in PCP signaling. Thus, the elegant regulation of polarity in the eye by graded Fj and Ds expression may represent only one of a number of ways to modulate the action of Ft. Further analysis of the mechanisms by which Ft and Ds regulate the pattern of Fz/PCP signaling will undoubtedly aid in the identification of these unknown signals and their mode of action.

This work was supported by a grant from the National Institute of General Medical Sciences (1RO1GM069923). I would like to thank Byron Miyazawa for help with injections, and Alana O'Reilly, Xuesong Zhao, Ami Okada and Jeff Axelrod for helpful discussion and comments on the manuscript. I would also like to thank P. Adler, F. Katz, D. Strutt and H. Sun for generously providing fly stocks.