The tissue polarity genes in Drosophila are required to coordinate cell polarity within the plane of the epidermis. Evidence to date suggests that these genes may encode components of a novel signal transduction pathway. Three of the genes, frizzled (fz), dishevelled (dsh), and prickle (pk) share a similar tissue polarity phenotype, suggesting that they function together in a single process. dsh is also known to function as a mediator of wingless (wg) signaling in a variety of developmental patterning processes in the fly. In this study, we make use of a fz transgene and a hypomorphic fz allele as genetic tools in an attempt to order these genes in a genetic hierarchy. Our results argue that dsh encodes a dosage sensitive component required for fz function and that it likely acts downstream of fz in the generation of tissue polarity. Our findings suggest that dsh may have a general role in signal transduction, perhaps as a component of a receptor complex.

Morphogenesis in multicellular organisms requires mechanisms to control the polarity and spatial arrangements of cells. In Drosophila, for example, the epidermis gives rise to polarized arrays of cuticular structures such as bristles, hairs (trichomes) and bracts, which point posteriorly on the thorax and abdomen and distally on appendages. The orientation of these structures reflects the coordination of cellular polarity within the plane of the underlying epidermis. As a model system for studying planar cell polarity, we have focused on the Drosophila wing because it is a relatively simple tissue with a well defined polarity pattern. Each cell on the wing blade elaborates a single distally oriented extension of the apical surface that gives rise to a cuticular hair. These hairs can be used as markers for cell polarity.

A number of genes required for coordinating cell polarity on the wing have been previously identified (see Adler, 1992 and Gubb, 1993 for review). Mutations in these ‘tissue polarity’ genes alter the polarity as well as the number of hairs produced by each cell, and these defects correlate with specific changes in the subcellular location of prehair initiation within pupal wing cells. This correlation led to the suggestion that the tissue polarity genes control hair polarity by restricting prehair initiation to the distal-most vertex of each cell (Wong and Adler, 1993). Mosaic studies indicated that this process involves intercellular signaling (Gubb and Garcia-Bellido, 1982; Vinson and Adler, 1987).

Based on analysis of adult and pupal wing phenotypes, six tissue polarity genes have been placed into three phenotypic groups which also represent epistasis groups (Wong and Adler, 1993). This analysis suggested a genetic hierarchy in which frizzled (fz), dishevelled (dsh), and prickle (pk) act upstream of fuzzy (fy) and inturned (in) which, in turn act upstream of multiple wing hair (mwh; Wong and Adler, 1993). In this study, we examined genetic interactions between fz, dsh, and pk in an attempt to order their function within the genetic hierarchy. The similar tissue polarity mutant phenotypes of these three genes suggests that their respective proteins may function together as components of a tissue polarity signaling pathway (Gubb and Garcia-Bellido, 1982; Wong and Adler, 1993).

The dsh gene is also known for its role as a member of the ‘wingless class’ of segment polarity genes. These genes are thought to encode components of a novel signaling pathway that functions in a variety of developmental patterning processes in the embryo and imaginal disks (see Klingensmith and Nusse, 1994 for review). In this context, Dsh is thought to mediate the transduction of a Wg signal from an unidentified Wg receptor (Klingensmith et al., 1994; Noordermeer et al., 1994; Seigfried et al., 1994; Theisen et al., 1994). dsh’s known role as a member of the wg signaling pathway raises the possibility that dsh could function in parallel to fz and/or pk to regulate tissue polarity.

We show here that dsh function is required to mediate fz activity and that it acts in a dosage sensitive manner. Thus, our results suggest that dsh function is not limited to transduction of Wg signal, but may instead play a more general role in planar signaling within epithelial tissues. In contrast, pk function is not required for fz activity and does not appear to function in a simple linear pathway with fz and dsh. Possible roles for fz, dsh and pk in tissue polarity signaling will be discussed.

Fly stocks

Most stocks are described by Lindsley and Zimm (1991). Several were obtained from the Drosophila stock centers at Indiana University and Bowling Green State University. Wild-type chromosomes were derived from either OreR, w, or yw stocks. The fzR53 allele has been described previously (Adler et al., 1987). The hsfz transgene used in these experiments is described in Krasnow and Adler (1994) where it is referred to as hsfzI. (The hsfz chromosome used in the dsh experiments contains 3 copies of the transgene and produces a particularly strong phenotype). For epistasis experiments, we constructed stocks that were homozygous for either a dsh1 or pk1 mutation and an hsfz transgene. In dsh dosage experiments, we used dsh1 and dshM20. In addition, we used two deficiencies (Df[1]v-N48 and Df[1]GA112), and a duplication (Dp[1;Y]BS−v+y+ Y) which cover the dsh locus. Dp[1;Y]BS−v+y+ Y is a compound chromosome in which a piece of the X chromosome has been translocated to the Y chromosome. Presumably, when this chromosome is carried in a male along with a wild-type X chromosome, dosage compensation will result in a 4X gene dosage of dsh+. For simplicity, we will refer to this chromosome in the text as Dpdsh+.

Mounting wings/microscopy

Wings were dehydrated in 100% ethanol and mounted in euparal. Light micrographs were produced using a confocal microscope (Molecular Dynamics) and processed using Adobe Photoshop.

Heat shock regimens

Flies were raised in vials at 25±1°C. Unless otherwise noted, staged pupae were transferred to glass vials and heat shocked for 1 hour in a water bath at 37-38°C. The sensitive period for the hsfz-late phenotype is between 33-35 hours AP. Therefore, to produce a strong hsfz-late phenotype we heat shocked at 34±0.5 hours AP. The sensitive period for the hsfz-early phenotype is between 0-30 hours after pupariation (AP). Within this period, the severity of the phenotype can be modulated by varying the time of heat shock. Earlier heat shocks produce correspondingly weaker phenotypes. Thus, to produce a weak hsfz-early phenotype we heat shocked at 20.5 hours AP and to produce a moderate to strong hsfz-early phenotype we heat shocked at 26 hours AP. See Krasnow and Adler (1994) for further details of hsfz overexpression phenotypes.

Quantitation of tissue polarity phenotypes

Both hair polarity and hair number phenotypes were scored on the dorsal surface in the distal C region of the wing (i.e., distal to the anterior crossvein and between the third and fourth longitudinal vein). Hair number phenotypes were scored by counting the number of multiple hair cells (MHCs), i.e., cells that produced 2 or more hairs. Hair polarity phenotypes were scored by measuring the fractional area of the wing exhibiting abnormal hair polarity, defined by a group of cells in which hair polarity differed from wild-type by greater than 45°. Areas with abnormal hair polarity were marked on a wing diagram and scanned using Adobe Photoshop. The fractional area of the marked regions was calculated by using the NIH image program to determine the number of marked pixels. All wings were scored blind.

Western analysis and immunostaining

Fz protein was detected essentially as described previously (Park et al., 1994). Wild-type and dsh samples were obtained from homozygous stocks. dsh; hsfz samples were obtained from a dsh/FM7; hsfz stock by selecting male larvae with y+ cuticle. Control male larvae were selected from an hsfz stock. For experiments involving hsfz, white prepupae (wpp) were collected and aged either at 25°C to 34 hours AP (for western analysis) or at 18°C to 68 hours AP (for immunostaining). Pupae were then heat shocked for 1 hour at 37°C and allowed to recover at 25°C for 2 hours before dissection. For western analysis of overexpressed Fz protein, whole pupal wings enclosed in the pupal sac were homogenized in 100 μl of sample buffer and loaded onto an SDS gel (4 wings per sample). For western analysis of endogenous Fz protein in wing imaginal disks, tissues were fractionated before loading to enrich for membrane proteins as previously described (Park et al., 1994). Immunostained pupal wings were examined under epiflourescence using a Zeiss fluorescence microscope and photographed using TMAX film (ASA 1600). Negatives were converted to digital images and processed using Adobe Photoshop.

Adult wing tissue polarity phenotypes

In mutants of each of the tissue polarity genes, the normal pattern of distally pointing wing hairs is replaced by a complex alternative pattern. These alternative patterns involve region specific alterations in hair polarity. In some wing regions, neighboring hairs are well aligned locally, but point in a nondistal direction, resulting in a swirling pattern. In other regions, local cell to cell alignment is lost, and hairs point in apparently random directions (see Gubb and Garcia-Bellido, 1982 and Wong and Adler, 1993 for details).

In addition to hair polarity defects, all of the tissue polarity mutants have an associated multiple hair cell (MHC) defect. Both the frequency of MHCs and the average number of hairs per cell differs dramatically depending on the genotype (Wong and Adler, 1993). For example, strong alleles of fz, dsh, and pk, result in fewer than 3% of cells in a central region of the wing forming more than a single hair, compared to about 70% of cells with strong alleles of fy and in (Adler et al., 1994; Wong and Adler, 1993; P. N. A., unpublished observations).

The hsfz-late phenotype requires dsh but not pk function

The overall similarity between the fz, dsh and pk phenotypes precluded the use of epistasis analysis to determine their order of function in a genetic hierarchy. Therefore, we took advantage of a genetic tool provided by a fz transgene. We showed previously that overexpression of fz via a heat shock promoter (hsfz) produces a range of tissue polarity phenotypes, the quality of which depends on the developmental timing of the heat shock (Krasnow and Adler, 1994). Heat shocks, given 6 or more hours before prehair initiation (i.e. at or before 30 hours AP), result in phenotypes that resemble those of fz, dsh, or pk mutants; heat shocks given just before prehair initiation (i.e. at 33-35 hours AP) result in distinct phenotypes that resemble those of fy or in mutants. In this paper, we will refer to these two distinct classes of heat shock induced phenotypes as hsfz-early and hsfz-late.

The hsfz-late phenotype can be distinguished from the fz, dsh, pk phenotypes by the dramatic difference in the number of MHCs (Krasnow and Adler, 1994). Thus, the hsfz-late phenotype provided a genetic tool that enabled us to do epistasis analysis with mutations in dsh, and pk. To do this, we constructed ‘double mutant’ stocks that were homozygous for a mutation in either dsh or pk in combination with an hsfz transgene. Staged pupae from these stocks were heat shocked alongside appropriate control pupae to induce a strong hsfz-late phenotype. Examination of the adult wings from these flies revealed that the high frequency of MHCs, characteristic of the hsfz-late phenotype, was dramatically suppressed in dsh; hsfz wings (Fig. 1). In some cases, the dsh; hsfz ‘double mutant’ wings were indistinguishable from those of dsh; + single mutants, indicating that a dsh mutation can completely block expression of the hsfz-late phenotype. Wings from flies with one functional copy of dsh+ (dsh/+; hsfz) had a phenotype that was intermediate in severity between those with no dsh+ (dsh; hsfz) and those with two wild type copies (+; hsfz).

Fig. 1.

A dsh, but not a pk mutation can suppress the hsfz-late MHC phenotype. (A-F) Light micrographs of wings from flies mutant for either dsh or pk alone or in combination with a hsfz transgene. The same region of the wing is shown in each panel (dorsal surface of the C region of the wing). Anterior is up and distal is to the right. MHCs can be distinguished from single hair cells by the closer spacing between hairs on the same surface of the wing. Note, the base of some hairs on the ventral surface of the wing are also visible at this focal plane. (G-L) Schematic diagrams of the wings shown in (A-F). Arrows indicate hair polarity. Hairs from MHCs are indicated in black. Hairs from single hair cells are indicated in grey. All flies were given a late heat shock. (A,G) hsfz control. Many cells in this region of the wing produce MHCs (typically two hairs per cell). (B,H) dsh/+; hsfz. These wings have an intermediate number of MHCs. (C,I) dsh; hsfz. Very few cells produce MHCs. These wings closely resemble those of dsh single mutants. (D,J) dsh. (E,K) pk; hsfz. There is a high frequency of MHC comparable to hsfz control wings from the same experiment (not shown). Note also, the hair polarity pattern is similar to that of pk. (F,L) pk. Very few MHCs are seen. Hairs in this region of the wing tend to point in an anterior direction.

Fig. 1.

A dsh, but not a pk mutation can suppress the hsfz-late MHC phenotype. (A-F) Light micrographs of wings from flies mutant for either dsh or pk alone or in combination with a hsfz transgene. The same region of the wing is shown in each panel (dorsal surface of the C region of the wing). Anterior is up and distal is to the right. MHCs can be distinguished from single hair cells by the closer spacing between hairs on the same surface of the wing. Note, the base of some hairs on the ventral surface of the wing are also visible at this focal plane. (G-L) Schematic diagrams of the wings shown in (A-F). Arrows indicate hair polarity. Hairs from MHCs are indicated in black. Hairs from single hair cells are indicated in grey. All flies were given a late heat shock. (A,G) hsfz control. Many cells in this region of the wing produce MHCs (typically two hairs per cell). (B,H) dsh/+; hsfz. These wings have an intermediate number of MHCs. (C,I) dsh; hsfz. Very few cells produce MHCs. These wings closely resemble those of dsh single mutants. (D,J) dsh. (E,K) pk; hsfz. There is a high frequency of MHC comparable to hsfz control wings from the same experiment (not shown). Note also, the hair polarity pattern is similar to that of pk. (F,L) pk. Very few MHCs are seen. Hairs in this region of the wing tend to point in an anterior direction.

To quantitate the phenotypes, we counted the number of MHCs in a well defined region of the wing (see Materials and Methods for details). These data confirmed that a dsh mutation can almost completely block expression of the hsfz-late phenotype (Fig. 2). In several repetitions of this experiment, we found that a dsh mutation always either blocked or strongly suppressed the hsfz-late phenotype. In one experiment, in which the +; hsfz control wings exhibited a particularly strong MHC phenotype (mean [s.e.]= 326 [9], n=10), the dsh; hsfz double mutants exhibited a moderate MHC phenotype (55 [11], n=8), indicating that the dsh mutation strongly suppressed but did not completely block the hsfz-late phenotype (not shown). We believe that this moderate MHC phenotype reflects residual function of the dsh1 allele used in these experiments (dsh1 is an adult viable, hypomorphic allele which produces a strong tissue polarity phenotype. All other dsh alleles are homozygous lethal due to its essential functions in embryogenesis).

Fig. 2.

Quantitation of MHC phenotypes in wings from flies mutant for either dsh or pk alone or in combination with an hsfz transgene. Wings from the experiments shown in Fig. 1 were scored by counting the number of MHCs in the dorsal C region of the wing (see Materials and Methods for details of scoring). The mean number of MHCs for each genotype is plotted. Bars indicate standard error, n= 4 to 8. (A) As in Fig. 1, there is a progressive reduction in the frequency of MHCs with reduced dsh+ gene dosage. Wings from dsh; hsfz double mutant flies had few MHCs, similar to those of dsh; + single mutants. (B) In contrast, wings from pk; hsfz flies had a high number of MHCs, similar to +; hsfz controls. Note that A and B represent scoring of wings from two separate experiments. The absolute level of MHCs varies in different experiments, depending on the copy number and strength of the particular transgene used, and the precise timing and temperature of the heat shock.

Fig. 2.

Quantitation of MHC phenotypes in wings from flies mutant for either dsh or pk alone or in combination with an hsfz transgene. Wings from the experiments shown in Fig. 1 were scored by counting the number of MHCs in the dorsal C region of the wing (see Materials and Methods for details of scoring). The mean number of MHCs for each genotype is plotted. Bars indicate standard error, n= 4 to 8. (A) As in Fig. 1, there is a progressive reduction in the frequency of MHCs with reduced dsh+ gene dosage. Wings from dsh; hsfz double mutant flies had few MHCs, similar to those of dsh; + single mutants. (B) In contrast, wings from pk; hsfz flies had a high number of MHCs, similar to +; hsfz controls. Note that A and B represent scoring of wings from two separate experiments. The absolute level of MHCs varies in different experiments, depending on the copy number and strength of the particular transgene used, and the precise timing and temperature of the heat shock.

In contrast to the clear epistasis of the dsh mutant phenotype, we found that a strong pk mutation (pk1) did not block expression of the hsfz-late MHC phenotype. That is, wings from pk; hsfz double mutants (Fig. 1E) displayed a high frequency of MHCs, similar to those of hsfz controls. In individual experiments, pk; hsfz wings had either more or fewer MHCs compared to control hsfz wings. However, this difference was usually not statistically significant and in no case was the difference dramatic, as was always seen in the experiments with dsh. In addition, we observed in some cases that the hair polarity pattern of the double mutants was reminiscent of the pattern in pk single mutants, albeit less extreme. For example, hairs in the C region of the wings of pk; hsfz double mutants pointed in an anterior direction as they do in pk single mutants (Fig. 1F). Thus, pk; hsfz double mutants exhibited a ‘mixed’ phenotype that included distinctive elements of each of the single mutants. We note that the hair polarity pattern of pk; fz double mutants also appears to be a mixture of the two single mutant patterns (Gubb and Garcia-Bellido, 1982; Wong and Adler, 1993).

fz activity is proportional to dsh gene dosage

The above experiments indicated that dsh function is required in a dosage sensitive manner for the expression of the hsfz-late phenotype. However, since overexpression phenotypes can sometimes be misleading about the normal function of a gene, we sought evidence that dsh is required for endogenous fz activity. Therefore, we determined the effect of varying dsh dosage on the severity of a hypomorphic fz allele, fzR53. To reduce dsh+ dosage in these experiments, we generated fzR53 flies that carried one copy of a dsh deficiency or null mutation (i.e., dsh/+ heterozygotes). To increase dsh+ dosage, we generated fzR53 flies that carried a dsh+ duplication (Dpdsh+). Since the fzR53 MHC phenotype is significantly weaker than that of the hsfz-late phenotype, we quantitated the phenotypes, by scoring both the number of MHCs and the fraction of the wing area having abnormal hair polarity (Krasnow and Adler, 1994; see Materials and Methods for details of scoring). In all cases, we obtained qualitatively similar results for both parameters.

The results of these experiments are shown in Figs 3 and 4. We found that a reduction of dsh+ gene dosage resulted in a dramatic enhancement of the fzR53 phenotype (Figs 3 and 4B,D). The severity of the dsh/+; fzR53 wing phenotype approached that of null mutations in fz or dsh. Conversely, we found that an increase in dsh+ dosage resulted in a suppression of the fzR53 phenotype (Fig. 4F). These data argue that dsh function is required for endogenous fz activity.

Fig. 3.

A dsh mutation dominantly suppresses the hsfz-early phenotype and enhances the fzR53 phenotype. Shown are light micrographs of wings from hsfz or fzR53 flies with either normal or reduced dsh dosage. The same region of the wing is shown in each panel (see Materials and Methods for details). Anterior is up and distal is to the right. Flies carrying an hsfz transgene were given an early heat shock to induce a moderate to strong hsfz-early phenotype. In heat shocked +; hsfz control flies, approximately 50% of the hairs have abnormal polarity (i.e. non distal). By contrast, in heat shocked dsh/+; hsfz flies, almost all hairs have normal distal polarity. +; fzR53 wings have a very weak phenotype and look almost wild type in this region of the wing. In dsh/+; fzR53 wings, most hairs have abnormal polarity. The phenotype resembles that of a strong fz or dsh mutant.

Fig. 3.

A dsh mutation dominantly suppresses the hsfz-early phenotype and enhances the fzR53 phenotype. Shown are light micrographs of wings from hsfz or fzR53 flies with either normal or reduced dsh dosage. The same region of the wing is shown in each panel (see Materials and Methods for details). Anterior is up and distal is to the right. Flies carrying an hsfz transgene were given an early heat shock to induce a moderate to strong hsfz-early phenotype. In heat shocked +; hsfz control flies, approximately 50% of the hairs have abnormal polarity (i.e. non distal). By contrast, in heat shocked dsh/+; hsfz flies, almost all hairs have normal distal polarity. +; fzR53 wings have a very weak phenotype and look almost wild type in this region of the wing. In dsh/+; fzR53 wings, most hairs have abnormal polarity. The phenotype resembles that of a strong fz or dsh mutant.

Fig. 4.

Quantitation of MHC and abnormal hair polarity phenotypes in hsfz-early or fzR53 flies with varying dsh+ gene dosage. Wings from the experiments shown in Fig. 3 were scored by counting the number of MHCs and by measuring the fraction of the wing area having abnormal hair polarity (see Materials and Methods for details of scoring). Bars indicate standard error, n=5. (A,B) The mean number of MHCs for flies with reduced dsh+ gene dosage. As expected, an increase in dsh+ dosage had the opposite effects, but the results were not statistically significant (not shown). (C-F) The mean percentage abnormal hair polarity for flies with reduced or increased dsh+ gene dosage. A and C show scores from flies that were heat shocked at 26 hours AP to induce a moderate to strong hsfz-early phenotype. E shows scores from flies that were heat shocked at 20.5 hours AP to induce a weak hsfz-early phenotype.

Fig. 4.

Quantitation of MHC and abnormal hair polarity phenotypes in hsfz-early or fzR53 flies with varying dsh+ gene dosage. Wings from the experiments shown in Fig. 3 were scored by counting the number of MHCs and by measuring the fraction of the wing area having abnormal hair polarity (see Materials and Methods for details of scoring). Bars indicate standard error, n=5. (A,B) The mean number of MHCs for flies with reduced dsh+ gene dosage. As expected, an increase in dsh+ dosage had the opposite effects, but the results were not statistically significant (not shown). (C-F) The mean percentage abnormal hair polarity for flies with reduced or increased dsh+ gene dosage. A and C show scores from flies that were heat shocked at 26 hours AP to induce a moderate to strong hsfz-early phenotype. E shows scores from flies that were heat shocked at 20.5 hours AP to induce a weak hsfz-early phenotype.

As a further test of the dosage relationship between fz and dsh we examined the effects of altering dsh dosage on the hsfzearly phenotype. The hsfz-early overexpression phenotype is distinct from the hsfz-late phenotype and paradoxically resembles a fz loss-of-function phenotype (Krasnow and Adler, 1994). To sensitize the hsfz-early phenotype to changes in dsh dosage, we modulated the severity of the hsfz-early phenotype by varying the timing of the heat shock (Krasnow and Adler,1994; see Materials and Methods for details). As shown in Figs 3 and 4A,C, we found that a reduction in dsh+ dosage resulted in a dramatic suppression of the hsfz-early phenotype. Con-versely, an increase in dsh+ dosage resulted in an enhancement of the hsfz-early phenotype (Fig. 4E). These results indicate that dsh function is required for production of the hsfz-early as well as the hsfz-late Fz overexpression phenotype. Moreover, the opposing effects of dsh dosage on the fzR53 loss of function phenotype versus the hsfz overexpression phenotypes argues that both hsfz overexpression phenotypes result from an increase in Fz activity. Taken together, the direction of the above genetic interactions indicates that fz activity is directly proportional to dsh gene dosage.

The effects of dsh gene dosage are not due to changes in developmental timing

Since similar experiments with several different alleles of dsh (see Materials and Methods), produced comparable results, the observed genetic interactions are almost certainly due to dsh dosage, and not to genetic background. In principle, variation of dsh gene dosage could affect the severity of hsfz phenotypes indirectly, by affecting developmental timing. Since the hsfz overexpression phenotypes are sensitive to the timing of heat shock, this is a valid concern. However, several points argue against this possibility. First, since the experiments involving fzR53 did not involve a heat shock, an effect on developmental timing cannot be the only explanation for our results. Second, to control for the possibility that a minor effect on developmental timing could delay the sensitive period for the hsfz-late MHC phenotype in dsh mutants, we examined wings from flies that were heat shocked at hourly time points between 33 and 38 hours AP (i.e., up to 5 hours beyond the onset of the sensitive period for control wings). However, in no case did dsh; hsfz wings exhibit an hsfz-late phenotype. Finally, to account for the nearly complete suppression of the hsfz-early overexpression phenotype, a 50% reduction in dsh dosage would have to have a dramatic effect on developmental timing. We estimate from our previous studies that a reduction in dsh dosage would either have to delay development by as much as 9 hours or advance it by as much as 8 hours (see Krasnow and Adler, 1994). Furthermore, a dsh duplication would have to affect developmental timing in the opposite direction. If a dsh mutation could affect developmental timing to this extent, we would expect to easily detect a shift in the time of prehair initiation in dsh mutants. However, previous studies indicated that a dsh mutation does not noticeably affect the time of prehair initiation, which begins at approximately 36 hours AP in all six tissue polarity mutants (Wong and Adler, 1993).

Fz protein is not altered in a dsh mutant

We considered the possibility that dsh function could modulate fz activity by affecting either the synthesis, stability, modification, or subcellular distribution of the Fz protein. We first tested whether suppression of the hsfz-late phenotype by a dsh mutation could result from an effect on the overexpressed Fz protein. In these experiments, staged pupae were subjected to a late heat shock (34 hours AP) and allowed to recover for 2 hours. Western blot analysis indicated that the apparent size and amount of Fz protein detected in hsfz versus dsh; hsfz pupal wings was roughly equivalent (Fig. 5A). Immunostaining further indicated that the subcellular distribution and intensity of Fz protein staining was also similar in the two genotypes (compare Fig. 5C and D). Thus, the ability of a dsh mutation to block the hsfz-late phenotype is not due to any obvious effect on the Fz protein.

Fig. 5.

A dsh mutation does not affect the size, amount, or distribution of Fz protein. (A,B) Western blot analysis of pupal wings and imaginal disks. (A) Overexpressed Fz protein detected 2 hours after heat shock of hsfz and dsh; hsfz pupal wings. Although a dsh mutation can block the phenotype of Fz overexpression, there is no apparent difference in the amount or size of Fz protein. (B) Endogenous Fz protein detected in wild-type is also comparable to that in dsh imaginal disks. Note, the magnitude of the signal is much lower in these samples and requires a longer exposure of the blot. (C,D) Immunostaining of pupal wings showing a similar distribution of overexpressed Fz protein detected 2 hours after heat shock of hsfz (C) and dsh; hsfz (D) pupae. In both cases, Fz protein is enriched at the cell periphery as it is in wild-type wings (not shown). The average intensity of staining was indistinguishable between the two genotypes.

Fig. 5.

A dsh mutation does not affect the size, amount, or distribution of Fz protein. (A,B) Western blot analysis of pupal wings and imaginal disks. (A) Overexpressed Fz protein detected 2 hours after heat shock of hsfz and dsh; hsfz pupal wings. Although a dsh mutation can block the phenotype of Fz overexpression, there is no apparent difference in the amount or size of Fz protein. (B) Endogenous Fz protein detected in wild-type is also comparable to that in dsh imaginal disks. Note, the magnitude of the signal is much lower in these samples and requires a longer exposure of the blot. (C,D) Immunostaining of pupal wings showing a similar distribution of overexpressed Fz protein detected 2 hours after heat shock of hsfz (C) and dsh; hsfz (D) pupae. In both cases, Fz protein is enriched at the cell periphery as it is in wild-type wings (not shown). The average intensity of staining was indistinguishable between the two genotypes.

To extend these experiments, we attempted to determine if a dsh mutation could affect the endogenous Fz protein. In these experiments we were limited by the low signal to noise ratio that current antibody reagents provide (Park et al., 1994). As expected from our previous results, western blot analysis indicated that the apparent size and amount of Fz protein was roughly equivalent in wild-type and dsh wing disks (Fig. 5B), arguing that a dsh mutation does not obviously affect the endogenous Fz protein. In addition, we attempted to compare endogenous Fz protein in wild-type and dsh pupal wings by immunostaining. Although we saw no differences in the intensity or subcellular distribution of Fz protein (not shown), the significance of this result is questionable, given the low signal to noise ratio in these preparations.

Models for tissue polarity gene function

Mosaic studies have revealed that fz has a complex role in the generation of planar tissue polarity in the wing (Gubb and Garcia-Bellido, 1982; Vinson and Adler, 1987). For example, genetically wild-type cells often display abnormal hair polarity when they are located near the distal border of a clone homozygous for a fz null mutation. The disruptive effect of fz mutant cells can extend to as many as ten cells from the clone border. This ‘domineering non-autonomy’ suggests that fz function is required for the generation or transmission of polarity information to neighboring cells. Although most fz alleles (including null alleles) exhibit this domineering non-autonomy in clones, a small subset of hypomorphic alleles do not (Vinson and Adler, 1987). With these alleles, all wild-type cells near the border of a clone have normal hair polarity. The behavior of these ‘cell autonomous’ alleles suggests that fz function is also required within the cells that produce it, for the intracellular response to polarity information. These mosaic studies have led to the suggestion that fz has a dual role in tissue polarity. It is required both for the intracellular response to polarity information and also for the intercellular transmission of polarity information to neighboring cells (Vinson and Adler, 1987). A variety of molecular studies suggest that fz encodes an integral membrane protein which shares many characteristics of G-coupled receptors (Vinson et al., 1989; Park et al., 1994). This single protein is necessary and sufficient for both fz functions (Krasnow and Adler, 1994).

One possible way to account for fz’s intracellular and intercellular roles in tissue polarity is to suggest that Fz functions biochemically both as a receptor as well as a ligand. The Drosophila Boss protein has a topology similar to that of Fz and this protein has been shown to function as a tethered ligand required for photoreceptor cell fate specification in the eye (Kramer et al., 1991). However, we favor an alternative model (Park et al., 1994; Krasnow and Adler, 1994; Vinson et al., 1989; Wong and Adler, 1993) in which Fz functions solely as a receptor. Thus, to account for fz’s two roles in tissue polarity, we propose that activation of Fz receptor leads to transduction of polarity information along two separate pathways (see Fig. 6). Transduction along an intercellular signaling pathway leads to the relay of polarity information to neighboring cells (nonautonomous function). Transduction along an intracellular response pathway leads to prehair initiation at the distal vertex of the cell (cell autonomous function). Previous genetic evidence suggests that this intracellular pathway may involve local inhibition of fy/in and mwh activity which restricts prehair initiation to the distal vertex (Krasnow and Adler, 1994; Wong and Adler, 1993).

Fig. 6.

A genetic model for a branched tissue polarity signal transduction pathway in the pupal wing. Grey circles represent a locally acting polarity signal which functions as a ligand for Fz. Activation of Fz receptor on the proximal side of the cell is transduced through both an intracellular response pathway and intercellular signaling pathway. Transduction through the intracellular response pathway leads to prehair initiation at the distal vertex of the cell (cell autonomous function). dsh may function as a component of this branch of the pathway by acting to antagonize fy/in activity at the distal vertex of the cell. Transduction through the intercellular signaling pathway results in a relay of the signal and its directional transmission (proximal to distal) to neighboring cells (non-autonomous function). pk may function as a component of this branch of the pathway.

Fig. 6.

A genetic model for a branched tissue polarity signal transduction pathway in the pupal wing. Grey circles represent a locally acting polarity signal which functions as a ligand for Fz. Activation of Fz receptor on the proximal side of the cell is transduced through both an intracellular response pathway and intercellular signaling pathway. Transduction through the intracellular response pathway leads to prehair initiation at the distal vertex of the cell (cell autonomous function). dsh may function as a component of this branch of the pathway by acting to antagonize fy/in activity at the distal vertex of the cell. Transduction through the intercellular signaling pathway results in a relay of the signal and its directional transmission (proximal to distal) to neighboring cells (non-autonomous function). pk may function as a component of this branch of the pathway.

The proposed branched signaling pathway is analogous to that for aggregation of Dictyostelium amoebae in response to cAMP. During aggregation, cAMP acts as a ligand for a G-coupled receptor (CAR1) which transduces the signal along at least two separate pathways, via distinct second messengers (see Devreotes, 1989 for review). Transduction along an intracellular response pathway leads to polarized cell migration. Transduction along a separate intercellular signaling pathway leads to release of cAMP, thereby relaying the signal to nearby cells. As discussed below, our results are consistent with this type of model and suggest that dsh likely encodes a component of the intracellular response pathway whereas pk may encode a component of the intercellular signaling pathway (Fig. 6).

dsh mediates fz function in tissue polarity

The striking similarity between the fz, dsh, and pk phenotypes suggested that these three genes might function together as components of a single tissue polarity signaling pathway. However, since dsh is known to function as a mediator of wg signaling (Klingensmith et al., 1994; Noordermeer et al., 1994; Seigfried et al., 1994; Theisen et al., 1994), it remained possible that fz, dsh, and pk could function in two or more parallel pathways. Our results support the idea that fz and dsh function together in a single linear pathway. For example, we found that a mutation in dsh can almost completely block the hsfz-late and hsfz-early phenotypes, indicating that dsh function is required for hsfz activity. Moreover, the strong genetic interactions between dsh and fzR53 imply that dsh function is also required for endogenous fz activity.

The simplest explanation for our results is that dsh functions downstream of fz in tissue polarity. Consistent with this idea, we found that a mutation in dsh has no obvious effect on the apparent size, amount, or subcellular distribution of Fz protein. Genetic mosaic evidence also supports the view that dsh functions downstream of fz. For example, most mutations in fz exhibit domineering non-autonomy in clones (Gubb and Garcia-Bellido, 1982; Vinson and Adler, 1987). In contrast, mutations in dsh only behave cell autonomously (Klingensmith et al., 1994; Theisen et al, 1994). Finally, molecular evidence favors the idea that dsh functions downstream of fz. For example, fz encodes a multipass integral membrane protein with both extracellular and intracellular domains. The topology is similar to that of G-coupled receptors and has led to the suggestion that Fz functions as a receptor for a polarity signal (Adler et al., 1990; Park et al., 1994; Vinson et al., 1989). In contrast, dsh encodes a novel intracellular protein which shares a region of sequence similarity to the discs-large homology region (DHR). This motif is found in several proteins that are associated with junctional complexes (Klingensmith et al., 1994; Theisen et al., 1994) and has recently been shown to be essential for dsh function in wg signal transduction (Yanagawa et al., 1995). Taken together, the evidence is consistent with a model in which Fz functions as a receptor for an intercellular polarity signal and Dsh functions downstream of Fz, as an intracellular mediator of the signal (see Fig. 6).

In principle, Fz could act either as an activator or an inhibitor of Dsh function. However, the direction of the dominant genetic interactions between fz and dsh argues that Fz likely activates Dsh. For example, we found that a reduction of dsh+ gene dosage enhances a fz loss-of-function phenotype (fzR53), but suppresses both types of fz gain-of-function phenotypes (hsfz-early and hsfz-late). Conversely, an increase of dsh dosage suppresses a fz loss-of-function phenotype, but enhances an hsfz-early gain-of-function phenotype. These results indicate that fz activity is proportional to dsh+ gene dosage, which suggests that Dsh functions as a positive mediator of Fz activity.

dsh function is not specific to wg signaling

Previous studies demonstrated that dsh functions as a mediator of wg activity (Noordermeer et al., 1994; Seigfried et al, 1994). Our results argue that dsh also functions as a mediator of fz activity. Assuming that Fz is not a receptor for Wg (see below), our results imply that Dsh does not function specifically in conjunction with a Wg receptor, nor does it add Wg specificity to a general receptor complex, as previously suggested (Klingensmith et al., 1994). Thus, we propose that Dsh has a general role as a downstream mediator of receptor activation in at least two distinct signal transduction pathways.

The robust genetic interactions between fz and dsh raise the possibility that the Fz and Dsh proteins may interact physically. Since the fz and dsh tissue polarity phenotypes are so similar, and we know of no other genes that function between fz and dsh, our results are consistent with the idea that Dsh may function as a component of a receptor complex, as has been suggested for its role in transduction of Wg signal (Klingensmith et al., 1994).

The proposed Fz and Wg signaling pathways have several interesting parallels. Both pathways presumably involve a novel signal transduction mechanism in which Dsh functions downstream of a receptor for a planar signal required for epidermal patterning. A dsh mutation can block the segment polarity phenotype caused by heat shock induction of Wg (Noordermeer et al, 1994) as well as the tissue polarity phenotype caused by heat shock induction of Fz.

These parallels between the Fz and Wg pathways led us to consider the possibility that Fz might function as a Wg receptor. Since null mutations in fz are viable, Fz cannot be the only Wg receptor. However, in principle, it could be a redundant one. This possibility could explain why a Wg receptor has not yet been identified, since mutations in both fz and a second gene would be required to produce a Wg-like mutant phenotype. Indeed, it has recently been proposed that Wg may have a role in tissue polarity (Theisen et al., 1994). Nevertheless, we think it is unlikely that Wg functions in the Fz signaling pathway, as we have not been able to detect any direct link. For example, we were unable to produce a tissue polarity phenotype by heat shock induction of an HSwg transgene (R. E. K., unpublished data). An intriguing possibility remains, however, that the Wg receptor could share structural features or sequence homology with Fz.

The role of pk in tissue polarity signaling

Genetic evidence indicates that pk is a member of a large, complex locus which includes the eye and leg specific tissue polarity gene spiny-leg (sple; Heitzler et al., 1993). Where pk fits into the putative Fz signaling pathway is not yet clear. Our results demonstrate that when subjected to a late heat shock, pk; hsfz double mutants produce the high frequency of MHCs characteristic of the hsfz-late phenotype. This result argues that pk, unlike dsh, does not function downstream of fz in a linear pathway. However, we hesitate to conclude that pk functions upstream of fz in a simple linear pathway because in some cases the hair polarity pattern of the pk; hsfz double mutants appeared to be a mixture of the two individual patterns.

Since mosaic analyses demonstrated that both pk and fz are required to transmit polarity information to neighboring cells (Gubb and Garcia-Bellido, 1982; Vinson and Adler, 1987), pk could function as a component of the signal relay branch of the proposed fz pathway (see Fig. 6A). At the genetic level, such a pathway would be circular, and genetic interactions would likely be complex. Moreover, the order of any two components would not be meaningful. It remains possible that pk could function on a separate independent pathway, parallel to that of fz. Determination of the molecular nature of the pk protein may help to sort out these possibilities.

We thank JingChun Liu for her superb technical assistance and Chris Turner for many helpful comments on the manuscript. We also thank N. Perrimon for providing the dshM20/FM7 and dshV26f36a/FM7 fly stocks and R. Nusse for providing the HSwg stock. This work was supported by a grant from the NIH (GM 37136). R. E. K. was supported by an NIH training grant in Developmental Biology (HDO 8812076).

Adler
,
P. N.
,
Charlton
,
J.
and
Vinson
,
C.
(
1987
).
Allelic variation at the frizzled locus of Drosophila
.
Dev. Genet
.
8
,
99
119
.
Adler
,
P. N.
,
Vinson
,
C.
Park
,
W. J.
,
Conover
,
S.
and
Klein
,
L.
(
1990
).
Molecular structure of frizzled, a Drosophila tissue polarity gene
.
Genetics
126
,
401
416
.
Adler
,
P. N.
(
1992
).
The genetic control of tissue polarity in Drosophila
.
BioEssays
14
,
735
741
.
Adler
,
P. N.
,
Charlton
,
J.
and
Park
,
W. J.
(
1994
).
The Drosophila tissue polarity gene inturned functions prior to wing hair morphogenesis in the regulation of hair polarity and number
.
Genetics
137
,
1
8
.
Devreotes
,
P.
(
1989
).
Dictyostelium discoideum: A model system for cell-cell interactions in development
.
Science
245
,
1054
1058
.
Gubb
,
D.
and
Garcia-Bellido
,
A.
(
1982
).
A genetic analysis of the determination of cuticular polarity during development in Drosophila melanogaster
.
J. Embryol. Exp. Morph
.
68
,
37
57
.
Gubb
,
D.
(
1993
).
Genes controlling cellular polarity in Drosophila
.
Development
Supplement
,
269
277
.
Heitzler
,
P.
,
Coulson
,
D.
,
Saenz-Robles
,
M-T.
,
Ashburner
,
M.
,
Roote
,
J.
,
Simpson
,
P.
and
Gubb
,
D.
(
1993
).
Genetic and cytogenetic analysis of the 43A-E region containing the segment polarity gene costa and the cellular polarity genes prickle and spiny-legs in Drosophila melanogaster
.
Genetics
135
,
105
115
.
Klingensmith
,
J.
and
Nusse
,
R.
(
1994
).
Signaling by wingless in Drosophila
.
Dev. Biol
.
166
,
396
414
.
Klingensmith
,
J.
,
Nusse
,
R.
and
Perrimon
,
N.
(
1994
).
The Drosophila segment polarity gene dishevelled encodes a novel protein required for response to the wingless signal
.
Genes Dev
.
8
,
118
130
.
Kramer
,
H.
,
Cagan
,
R. L.
and
Zipursky
,
S. L.
(
1991
).
Interaction of bride of sevenless membrane bound ligand and the sevenless tyrosine-kinase receptor
.
Nature
352
,
207
212
.
Krasnow
,
R. E.
and
Adler
,
P. N.
(
1994
).
A single frizzled protein has a dual role in tissue polarity
.
Development
120
,
1883
1893
.
Lindsley
,
D. L.
and
Zimm
,
G. G.
(
1992
).
The genome of Drosophila melanogaster
.
New York
:
Academic Press
.
Noordermeer
,
J.
,
Klingensmith
,
J.
,
Perrimon
,
N.
and
Nusse
,
R.
(
1994
).
dishevelled and armadillo act in the wingless signaling pathway in Drosophila
.
Nature
367
,
80
83
.
Park
,
W. J.
,
Liu
,
J.
and
Adler
,
P. N.
(
1994
).
The frizzled gene of Drosophila encodes a membrane protein with an odd number of transmembrane domains
.
Mech. Dev
.
45
,
127
137
.
Seigfried
,
E.
,
Wilder
,
E. L.
and
Perrimon
,
N.
(
1994
).
Components of wingless signaling in Drosophila
.
Nature
367
,
76
79
.
Theisen
,
H.
,
Purcell
,
J.
,
Bennett
,
M.
,
Kansagara
,
D.
,
Syed
,
A.
and
Marsh
,
J. L.
(
1994
).
dishevelled is required during wingless signaling to establish both cell polarity and cell identity
.
Development
120
,
347
360
.
Vinson
,
C.
and
Adler
,
P. N.
(
1987
).
Directional non-cell autonomy and the transmission of polarity information by the frizzled gene of Drosophila
.
Nature
329
,
549
551
.
Vinson
,
C. R.
,
Conover
,
S.
and
Adler
,
P. N.
(
1989
).
A Drosophila tissue polarity locus encodes a protein containing seven potential transmembrane domains
.
Nature
338
,
263
264
.
Yanagawa
,
S.
,
van Leeuwen
,
F.
,
Wodarz
,
A.
,
Klingensmith
,
J.
and
Nusse
,
R.
(
1995
).
The Dishevelled protein is modified by Wingless signaling in Drosophila
.
Genes Dev
.
9
,
1087
1097
.
Wong
,
L. L.
and
Adler
,
P. N.
(
1993
).
Tissue polarity genes regulate the subcellular location for prehair assembly in the pupal wing
.
J. Cell Biol
.
123
,
209
221
.