Abstract

Mutations in the Van Gogh gene result in the altered polarity of adult Drosophila cuticular structures. On the wing, Van Gogh mutations cause an altered polarity pattern that is typical of mutations that inactivate the frizzled signaling/signal transduction pathway. The phenotype however, differs from those seen previously, as the number of wing cells forming more than one hair is intermediate between that seen previously for typical frizzled-like or inturned-like mutations. Consistent with Van Gogh being involved in the function of the frizzled signaling/signal transduction pathway, Van Gogh mutations show strong interactions with mutations in frizzled and prickle. Mitotic clones of Van Gogh display domineering cell nonautonomy. In contrast to frizzled clones, Van Gogh clones alter the polarity of cells proximal (and in part anterior and posterior) but not distal to the clone. In further contrast to frizzled clones, Van Gogh clones cause neighboring wild-type hairs to point away from rather than toward the clone. This anti-frizzled type of domineering nonautonomy and the strong genetic interactions seen between frizzled and Van Gogh suggested the possibility that Van Gogh was required for the noncell autonomous function of frizzled. As a test of this possibility we induced frizzled clones in a Van Gogh mutant background and Van Gogh clones in a frizzled mutant background. In both cases the domineering nonautonomy was suppressed consistent with Van Gogh being essential for frizzled signaling.

THE cuticular surface of Drosophila is decorated with large numbers of polarized structures such as hairs, bristles and ommatidia. In any body region these structures are aligned in parallel, giving the region a “tissue polarity.” In recent years the genetic basis for the development of tissue polarity in the wing, eye and abdomen has been studied in some depth (Adler 1992; Gubb 1993; Zhenget al. 1995; Kopp and Duncan 1997; Struhlet al. 1997; Struttet al. 1997). We have used the wing as a model system because of the simple, flat structure of the tissue and the ease of examining the array of distally pointing hairs that decorate it. Almost all pupal wing cells form a single microvillus-like prehair that gives rise to the adult cuticular hair (Wong and Adler 1993). The prehairs are formed at the cell periphery in the vicinity of the distal-most region of the cell and extend distally as they grow, leading to the distally pointing adult cuticular hair (Wong and Adler 1993). Hair polarity is tightly correlated with the subcellular location for prehair initiation. Mutations that cause nondistal polarity appear to do so via altering the subcellular location for prehair initiation (Wong and Adler 1993). Previous studies have suggested that most, if not all, known tissue polarity genes are members of the frizzled (fz) signaling/signal transduction pathway that controls hair polarity by regulating the subcellular location for prehair initiation (Wong and Adler 1993). This pathway contains both cell autonomously and cell nonautonomously acting genes (Vinson and Adler 1987; Wong and Adler 1993). Several aspects of the tissue polarity phenotype have been used to characterize tissue polarity mutants and genes. The cellular phenotype (number of hairs per cell and the subcellular location for prehair initiation) allows mutants to be placed into three phenotypic groups that also represent epistasis groups [the frizzled (fz)-like genes, the inturned (in)-like genes and the multiple wing hair (mwh) gene; Wong and Adler 1993]. An alternative way to classify tissue polarity genes is based on the abnormal mutant wing polarity patterns that are stereotypic for individual genes (Gubb and Garcia-Bellido 1982; Wong and Adler 1993). Most tissue polarity mutations result in a similar, albeit not identical, pattern that we refer to as the fz/in-like polarity pattern. Mutations that inactivate the fz signaling/signal transduction pathway result in this pattern. Mutations that alter the anatomical direction of fz signaling cause unique and different patterns (Adleret al. 1998; R. E. Krasnow and P. N. Adler, unpublished results).

We recently carried out a large mutant screen designed to identify and recover new tissue polarity mutations. In addition to recovering recessive mutations as the screen was designed to do, we also identified and recovered a number of dominant tissue polarity mutations. One gene identified in this screen because of a dominant mutant phenotype, Van Gogh (Vang), is the subject of this paper.

The first Vang mutation was identified on the basis of a region of hair swirling in the C″ cell of the wing (this is the region of the wing that lies between the third and fourth veins proximal to the anterior cross vein). Many, but not all Vang alleles display this dominant phenotype. Homozygous Vang mutant wings show a tissue polarity phenotype that in terms of the cellular phenotype falls in between typical frizzled-like and inturned-like mutant phenotypes (Wong and Adler 1993; Krasnow and Adler 1994), suggesting that this gene might have a unique function in tissue polarity. Vang mutations cause a fz/in-like polarity pattern suggesting that mutations in Vang inactivate the fz pathway.

We have used the weak dominant phenotype to look for interactions with other tissue polarity genes. The two most compelling were the dominant enhancement of this phenotype by frizzled mutations and the dominant suppression of this phenotype by prickle (pk) mutations. Other genetic interactions were also seen between Vang and both fz and pk, suggesting a close functional relationship between these genes.

Mutations in fz are notable for the directional domineering cell nonautonomy shown by most alleles (Vinson and Adler 1987). Cells distal (and in part posterior and anterior) to a fz clone show altered hair polarity, with the neighboring wild-type hairs tending to point toward the clone (Adleret al. 1997). Marked mitotic clones of two independent Vang alleles were generated and found to show a remarkable “anti-fz” phenotype. That is, neighboring wild-type cells located proximal (and anterior and posterior) but not distal to the clone showed altered hair polarity. These wild-type hairs tended to point away from the clone—the opposite of what is seen for fz. To test if these opposite phenotypes could be because of an interaction between these two genes, we generated fz clones in a Vang mutant background and Vang clones in a fz mutant background. In both cases the cell nonautonomy of fz and Vang, respectively, were substantially suppressed. The data suggest that Vang is essential for fz intercellular signaling.

MATERIALS AND METHODS

Fly culture and strains: Flies were grown on standard media at 25° (unless stated otherwise). Many mutant- and deficiency-containing stocks were obtained from the stock centers at Indiana University, Bowling Green State University, and Umeå University. Several important deficiency chromosomes were kindly provided by the Konev laboratory. We carried out a large-scale mutant screen that utilizes FLP/FRT technology (Golic and Lindquist 1989; Xu and Rubin 1993) to recover mutations that cause altered hair polarity. The details of this screen will be presented elsewhere (P. N. Adler, J. Charlton and J. Liu, unpublished results). Briefly, FRT-carrying flies were mutagenized with EMS, crossed to hs-flp;FRT-carrying flies and clones induced in the F1 progeny via heat shocking larvae. The adult F1 flies were anesthetized under CO2, and one wing was removed without killing the fly. The wing was examined under a compound microscope and flies that had clones with altered hair polarity, number, or morphology were saved and bred to recover the mutation. In these screens we also recovered dominant mutations. The isolation of Vang mutations because of their dominant phenotype resulted in the start of the experiments reported here. Additional Vang alleles were isolated after EMS or γ-ray mutagenesis by their failure to complement an existing Vang allele. The gene was named Van Gogh because the swirling wing hair patterns bring to mind the swirling brush strokes this artist used in some of his paintings.

Cytological procedures: To examine the process of hair morphogenesis pupal wings were dissected in PBS + 4% paraformaldehyde, stained with a fluorescent phalloidin (that binds to F-actin; Wong and Adler 1993), and examined by confocal microscopy using a Molecular Dynamics (Sunnyvale, CA) confocal microscope.

Generation of genetic mosaics: Several types of mosaic experiments were carried out. In all we used the FLP/FRT system to generate clones of interest. For example, to determine if Vang acted in a cell in an autonomous or nonautonomous fashion we constructed larvae that were w hsflp; FRT42 Vang A3 kojakVB13/FRT42. These were heat shocked at 38° for ½ hr to induce the expression of the hs-flp gene and recombination at the FRT site. The subsequent clone could be recognized because of being homozygous for the recessive mutation kojak (koj). This mutation results in some cells forming multiple, split, and shortened hairs and others no hair (i.e., a bald cell). In other experiments we have found that koj is cell autonomous. The koj alleles used were isolated in our FLP/FRT screen. We used a new hair morphology marker starburst (strb) to mark fz clones (Parket al. 1996). In other experiments we also used pawn (pwn) as a cell marker for 2R and flare (flr) as a cell marker for 3L.

Scoring of mutant wings: Wings from relevant flies were mounted in Euparal (Asco Labs) and examined under bright field microscopy. As part of the analysis we often made drawings of individual wings that showed the abnormal polarity pattern of wing hairs on the dorsal surface of the wing. The pattern shown is of an individual wing; however, at least five other wings of that genotype were examined to insure that the wing drawn was typical for the genotype. In previous studies we have used several different quantitative assays to characterize tissue polarity phenotypes (e.g., number of multiple hair cells). These are described in detail elsewhere (Wong and Adler 1993; Krasnow and Adler 1994; Adleret al. 1994; Parket al. 1996).

RESULTS

Isolation of Vang: Our two original Vang alleles were recovered because of a dominant phenotype—a swirl in the wing hair pattern in the C″ region of the wing (this is the region that lies between the third and fourth veins proximal to the proximal cross vein; Figure 1B—compare to the wild-type pattern in this region; Figure 1A). Eight additional alleles were isolated via screening for a lack of complementation with the original allele. Flies homozygous for Vang alleles show a tissue polarity bristle phenotype on the wing, thorax, legs and abdomen. On the abdomen, bristles point almost orthogonally to the midline instead of posteriorly. The tarsus joints are often duplicated as is typical for tissue polarity mutants (Heldet al. 1986). The flies also have rough eyes, which we suspect is because of a tissue polarity effect (Zhenget al. 1995). Vang was mapped to meiotic map position 60 on 2R. It was mapped to cytological interval 45AB on the basis of being uncovered by Df(2R)NP4 (44F; 45B) and Df(2R)w45-30N (45A; 45E). The presence of either of these deficiencies is simply noted by Df in the text.

Figure 1.

All micrographs are of the dorsal surface of the wing. A shows the C″ region of a wild-type Oregon R wing. B shows the C″ region of a VangTBS42/+ wing. Note the swirling hairs in this region. C–F show the middle part of the C cell (distal to the posterior cross vein) in: (C) Oregon R, (D) VangTBS42/VangTBS42, (E) Vang A3/VangA3, and (F) Vang 4014/Vang4014 wings. Arrows are in the local direction of polarity. Note that Vang TBS42/VangTBS42 has a slightly stronger phenotype than VangA3/VangA3 and that Vang4014/Vang4014 has the strongest polarity phenotype, but very few multiple hair cells. In all wing figures proximal is to the left and distal to the right.

Several Vang alleles, e.g., VangTBS42 (see Table 1), showed complete penetrance for the dominant C″ swirl phenotype (Figure 1B). For these alleles (and indeed for all Vang alleles that showed the dominant C″ phenotype) the phenotype was stronger in flies raised at 29° than 18°. Deficiencies for the region showed no dominant C″ phenotype at 18° and a weak and incompletely penetrant one at 29°. By this genetic criterion we conclude that Vang alleles such as VangTBS42 that show complete penetrance for this dominant phenotype have at least some antimorphic character. We further conclude that the temperature sensitivity of the dominant phenotype is not because of temperature-sensitive mutant proteins. Several alleles, such as VangA3, showed a weak and incompletely penetrant dominant C″ phenotype. This was similar to what was seen with deficiencies for the region; hence by this genetic test (and for this phenotype) VangA3 appears to be amorphic and it may represent a null allele. Several Vang alleles did not show any evidence of a dominant phenotype and are likely hypomorphic alleles.

Phenotypic characterization of Vang: Homozygotes for 9 of the 10 Vang alleles showed a similar tissue polarity phenotype that differed only in severity (Figure 1, D–F). Consistent with the suggestion above that Vang A3 is an amorphic allele, Vang A3/Vang A3 and Vang A3/Df flies had similar phenotypes. Consistent with the suggestion that VangTBS42 is an antimorphic allele, homozygous VangTBS42 flies appeared to have a stronger wing phenotype than VangTBS42/Df flies. However, for other alleles the strength of the dominant phenotype did not always predict the strength of the homozygous phenotype. For example, based on the strong dominant C″ phenotype of Vang A5/+ wings, we considered Vang A5 to be an antimorphic allele. However, Vang A5/Df wings showed a phenotype that appeared weaker than Vang A3/Df. Thus, we consider this allele to be antimorphic for the dominant C″ phenotype but hypomorphic for the recessive wing hair polarity phenotype. Thus, our data argue that the genetics of Vang are complex (further evidence for complexity is found in observations on Vang4014 described below).

View this table:
TABLE 1

Van Gogh alleles

To assess in a quantitative way if Vang alleles had a fz-like (few multiple hair cells) or in-like (many multiple hair cells) cellular phenotype, we determined and plotted the fraction of the dorsal C cell with abnormal hair polarity and the number of multiple hair cells in this region. Previously we found that fz-like and in-like genes were easily distinguished by this assay (Krasnow and Adler 1994; Adleret al. 1998). All except one Vang allele fell between the patterns seen for fz-like and in-like genes as described previously (Figure 2). This included alleles that, on the basis of their dominant phenotypes, are antimorphic (VangTBS42) or amorphic (Vang A3). The one Vang allele that appeared different from the others (Vang4014) fell into the fz-like group, as it resulted in very few multiple hair cells although it had a more severe polarity phenotype than any other Vang allele (Figure 1F). As a further test that Vang4014 was a Vang allele (and not a mutation in a second gene that interacted with Vang), we generated Vang4014/VangTBS42 females and crossed them to VangTBS42/CyO males and examined the progeny for possible wild-type recombinants. None was found among more than 340 straight-winged progeny, consistent with Vang4014 being an unusual Vang allele. We conclude that Vang cannot be considered a typical fz-like or in-like gene with respect to this cellular phenotype. In examining Vang wings we observed dramatic local variation in the fraction of multiple hair cells. While local variation in the fraction of multiple hair cells is seen with other tissue polarity mutants, it appeared more extreme for Vang than other genes.

Figure 2.

A plot of the number of multiple hair cells in the dorsal C cell distal to the anterior cross vein as a function of the fraction of this region of the wing that has abnormal polarity. All points except the Vang points are taken from Krasnow and Adler (1994) and Adler et al. (1998). The darker symbols are stronger alleles and the lightest symbols the weakest alleles of these genotypes. fz-gofI stands for the early fz gain-of-function phenotype and fz-gof II for the late fz gain-of-function phenotype (Krasnow and Adler 1994). The five Vang genotypes are noted in the figure.

We examined phalloidin-stained Vang pupal wings and found that prehairs were formed either in the central regions of the apical surface or at alternative locations along the apical cell periphery of the wing epidermal cells (data not shown). This pattern is typical of the phenotypes of mutations in fz-like genes (Wong and Adler 1993).

An alternative scheme for categorizing tissue polarity mutants is the overall abnormal polarity pattern (Adleret al. 1998). Mutations in these genes cause stereotypic alterations to hair polarity across the wing. Most tissue polarity mutations fall into the fz/in polarity group. While mutations in these genes do not produce identical abnormal polarity patterns, there is substantial similarity between the mutant patterns. For example, in all the genes in this group hairs in the D and E cell tend to point toward the wing margin and away from the anterior/posterior compartment boundary. This stands in clear contrast to the unique and quite different patterns seen in ds and pk mutants (Gubb and Garcia-Bellido 1982; Wong and Adler 1993; Adleret al. 1998). By this criteria all ten of our Vang alleles fall into the fz/in polarity group (Figure 3).

Double mutant analysis and gene interactions: We constructed double mutants of Vang with fz, dsh (dishevelled), in, pk, and mwh. In these experiments we used VangTBS42, Vang A3, and Vang A5 and obtained similar results with all. The double mutants of Vang with fz, dsh, in, and mwh all showed the general hair polarity pattern typical of the fz/in group, to which these genes belong (Figure 4, A and B). The pk Vang double mutants also had a fz/in-like polarity pattern (Figure 4, C and D) (although perhaps less severe than Vang single mutants). Thus, by the polarity pattern criterion Vang is epistatic to pk.

Figure 3.

Shown are drawings of the wing hair polarity pattern on the dorsal surface of a wing of the indicated genotypes. The drawing is of an individual wing, although at least five wings were examined to insure that the drawing represents a typical wing. The A, B, C, D, and E cells of the wing are designated in the Oregon R panel. The filled regions represent regions where neighboring hairs do not show a common polarity (i.e., hair polarity is close to random in these regions).

Figure 4.

All micrographs are of the dorsal surface of the wing in the middle C region as in Figure 1. (A) dsh1/dsh1; (B) dsh1/dsh1; VangTBS42/VangTBS42; (C) pk1/pk1; (D) pk1 Vang TBS42/pk1 Vang TBS42; (E) fzR53/fzR53; and (F) VangTBS42/+;fzR53/fzR53. Note the dsh1/dsh1; VangTBS42/VangTBS42 wing (B) shows the low number of multiple hair cells typical of dsh and not the higher number seen in most Vang mutants (compare to Figure 1D). The pk1 Vang TBS42/pk1 Vang TBS42 wing has a Vang-like polarity pattern and not the pk polarity pattern (C vs. D and compare to Figure 1D) and an increased number of multiple hair cells. The weak fz allele fzR53/fzR53 is strongly enhanced by a single dose of VangTBS42 (compare E and F).

On the basis of casual observation it appeared that the multiple hair cell phenotypes of dsh, in, and mwh were epistatic to Vang. We confirmed this by counting the fraction of multiple hair cells in a 20 × 5 cell region just anterior to the posterior cross vein (Wong and Adler 1993). Thus, all of the strictly cell autonomously acting tissue polarity genes tested appeared to be epistatic to Vang. Several strong genetic interactions were seen between pk and Vang and between fz and Vang. These are discussed in more detail below.

In other experiments we used the dominant C″ swirl phenotype of Vang as a sensitized genetic background to look for dominant interactions of Vang and other tissue polarity genes. In these experiments we principally used VangTBS42, as the relatively strong C″ phenotype displayed by this antimorphic allele allowed us to look for enhancement or suppression in a single cross. We did not see any clear-cut interaction between mutations in dsh, in, fuzzy (fy), fritz, starry night, or mwh and the Vang-dominant phenotype. We also failed to see any interaction between the Vang-dominant phenotype and deficiencies for the Wnt-encoding genes wingless (wg) (and Wnt4), Wnt2, or Wnt3.

The interaction of Vang and pk: We found that loss-of-function mutations in prickle (pk) acted as dominant suppressors of the Vang-dominant C″ phenotype, suggesting these two genes act in an antagonistic fashion. Several different pk alleles (pk1, pk JJ13, Df(2R)pk78s) and several different Vang alleles (VangTBS42, Vang14-11, Vang11-3, VangA3) were used, and this interaction was seen in all combinations tested.

pk is a slightly haplo-insufficient gene. A deficiency for pk (and some pk point mutants) shows a weak, partially penetrant dominant tissue polarity phenotype. This is seen as a swirling of hairs in the D and E cells just distal to the posterior cross vein. Several, but not all, Vang alleles acted as enhancers of this haplo-insufficiency of pk. For example, Vang A3 and VangTBS42, but not Vang11-3, acted as dominant enhancers of the dominant phenotype of Df(2R)pk78s. The difference between VangA3 and Vang11-3 is interesting as by other tests both of these alleles were classified as amorphic (Table 1). We also found that Vang alleles acted as dominant enhancers of the tissue polarity phenotype seen in flies that carry a single copy of the dominant antimorphic pk allele pkTBJ21 (R. E. Krasnow and P. N. Adler, unpublished results). These enhancements were unexpected as this interaction is in the opposite direction to the suppression of the Vang-dominant C″ phenotype by pk mutations. Thus, the interaction between Vang and pk appears complex.

As noted earlier, double mutants of pk and Vang showed the fz/in polarity pattern. Interestingly, the number of multiple hair cells in the Vang; pk double-mutant wings was increased above that seen in either single mutant (e.g., in our standard test region just anterior to the posterior cross vein we found the following: VangTBS42, 1.39 hairs/cell; pk1, 1.01 hairs/cell; and pk1 VangTbs42, 1.7 hairs/cell). In previous experiments we saw no evidence for any additive or synergistic interactions for this phenotype (Wong and Adler 1993).

The interaction of Vang and fz is complex: Several different fz mutations (Df(3L)fzD21; In(3L)fzK21, fzR52) were found to act as dominant enhancers of the Vang-dominant C″ phenotype associated with Vang TBS42 and Vang A5. To determine if Vang mutations could enhance fz mutations we utilized the weak fz allele fzR53, which we have previously found to be a sensitive genetic background for detecting genetic interactions (Krasnowet al. 1995). We found that Vang alleles considered to be either antimorphic or amorphic (VangTBS42 and Vang A3) both acted as strong dominant enhancers of fzR53 (Figure 4, E and F). From these experiments it appeared that Vang and fz interacted in a positive fashion.

We constructed and examined the wings of several allelic combinations of Vang; fz double mutants. In VangTBS42; fz1, VangA3; fz1, Vang A5; fz1, VangTBS42; fzR54; Vang A3; fzR54/fzK21, and VangTBS42; fzK21/fzR54 wings the fz/in type of polarity pattern was seen, although surprisingly it generally appeared slightly less abnormal than in either single mutant. The number of multiple hair cells was reduced from that seen in the Vang single mutants (e.g., Vang A3, 1.31 hairs per cell; fzR54, 1.02 hairs per cell; and Vang A3;fzR54, 1.13 hairs per cell), although this was rather variable for the VangTBS42; fz1 flies both with respect to different individuals and different regions of the wing. Thus, the double mutant studies on fz and Vang, which gave if anything less severe mutant phenotypes than the single mutants, stand in contrast to those described above where we saw enhancement of mutant phenotypes.

We used Western blot analysis to examine the level of Fz protein in wild-type and Vang wing discs. No differences were found, arguing that Vang does not regulate fz expression.

Vang is not required for the transduction of the fz signal: The overexpression of fz just before prehair initiation causes the formation of large numbers of multiple hair cells that are a phenocopy of the in-like mutations (Krasnow and Adler 1994). This is consistent with our suggestion that a consequence of fz signal transduction is the inhibition of the activity of the products of the inturned-like genes in the vicinity of the distal vertex (Wong and Adler 1993). Evidence that this in phenocopy results from Fz signal transduction antagonizing In/Fy comes from experiments in which we found that the mild overexpression of fz (not enough to cause a substantial phenotype by itself) acts as a strong enhancer of weak alleles of in and fy (R. E. Krasnow and P. N. Adler, unpublished results). We have previously used the ability of fz overexpression to phenocopy in as a test to identify genes that are downstream of and required for the transduction of the fz signal (Krasnowet al. 1995). In these experiments we found that the cell autonomously acting dsh gene (Klingensmithet al. 1994; Theisenet al. 1994), but not pk or dachsous, was required for this phenocopy (Krasnowet al. 1995; Adleret al. 1998). To determine if Vang was required for the transduction of the fz signal we constructed Vang; hs-fz flies and induced fz expression just before prehair initiation. We found that the Vang mutation did not block the ability of fz overexpression to induce cells to form multiple hairs (Table 2). Indeed, if anything it appeared to enhance the ability of fz overexpression to induce multiple hair cells.

Vang is required for the ability of a gradient of fz expression to repolarize wing hairs: We have found that the induction of a gradient of fz expression, with its high point near the distal tip of the wing, results in a reversal of the normal distal polarity of hairs in this region of the wing (Adleret al. 1997). Because a similar induction of expression of dsh does not produce a region of reversed polarity, it seems likely that polarity reversal requires the cell nonautonomous function of fz. In a dsh mutant background the induction of a gradient of fz expression does not produce a region of reversed polarity; hence, a functional fz signal transduction pathway also appears to be needed. To determine if Vang function was required for a gradient of fz expression to cause a local reversal of wing hair polarity, we induced such a gradient of expression by “distal waxing” of Vang;hs-fz pupae (Adleret al. 1997). This did not result in a local region of proximal polarity (Table 3); hence, we concluded that Vang was required either for the cell nonautonomous function of fz or the transduction of the fz signal. Because the experiments described above argued that Vang was not required for the transduction of the fz signal, we were left with the suggestion that Vang is required for the cell nonautonomous function of fz.

Vang acts cell nonautonomously: In early experiments we found that when we induced unmarked Vang clones we saw a tissue polarity phenotype in wings. Thus, it was clear that the Vang phenotype would not be completely rescued by neighboring wild-type cells. We generated Vang koj clones for two Vang alleles (VangTBS42 and Vang A3) to determine if clones of Vang cells would disrupt the polarity of neighboring wild-type cells as do clones of fz (Vinson and Adler 1987; cells homozygous for koj produce either no hair or shortened multiple hairs). For both alleles we found that Vang clones resulted in regions of surrounding wild-type cells with abnormal polarity (102 of 103 VangTBS42 clones and 108 of 111 Vang A3 clones showed the domineering nonautonomy). This domineering nonautonomy was usually seen over less than half of the clone border. In striking contrast to what we have seen with fz, the domineering nonautonomy of Vang clones was seen in wild-type cells along the proximal (and in part anterior and posterior), but not the distal border of the clone (Figure 5, A–C). Further, the wild-type hairs showing abnormal polarity tended to point away from the clone border, rather than toward it as is seen for fz clones (Adleret al. 1997).

View this table:
TABLE 2

Vang is not required for the transduction of the frizzled signal

In these clone experiments we generated not only a Vang/Vang clone, but also a +/+ twin spot. The cells in this +/+ twin spot might be expected to have a higher level of Vang activity than the surrounding Vang/+ cells. To determine if these cells influenced the domineering nonautonomy of Vang we generated Vang A3 koj and pwn twin spots. The pwn twin spot cells, which can be recognized because of their thin wispy hairs, would have two copies of the wild-type Vang gene and hence potentially higher Vang activity than the surrounding cells. We saw no consistent pattern to the location of the pwn/pwn twin spot as compared to the Vang koj/Vang koj clone or the domineering nonautonomy (we examined 29 twin spots). Indeed, it was easy to find examples where the twin spot was far removed from the wild-type cells showing the domineering nonautonomy (Figure 5A). Hence, we concluded that the presence of cells with two wild-type Vang genes did not play an important role in the domineering nonautonomy of Vang clones.

We have carried out a similar twin spot analysis for fz (using fzR52 strb and flr twin spots) (Figure 5D). Here we also found no consistent pattern to the location of the fzR52 strb and flr twin spots (we examined 23 twin spots), leading to the conclusion that the twin spot cells with two doses of fz were not important for generating the distal domineering nonautonomy of fz (Vinson and Adler 1987).

Vang and fz mutations suppress the domineering cell nonautonomy of fz and Vang clones, respectively: The opposite domineering nonautonomy of fz and Vang clones, and the genetic interactions described above suggested the possibility that fz was involved in the domineering nonautonomy of Vang (and vice versa). To determine if fz activity was required for the domineering nonautonomy of Vang we generated Vang A3 koj clones in a fz mutant background. We scored clones for nonautonomy in those regions (i.e., peripheral as opposed to central regions of the wing) where the polarity of hairs in a fz mutant was consistent even if abnormal. For example, hairs consistently point toward the margin in the peripheral regions of the D and E cells (see Figure 3). Although fz and Vang mutations do not produce strong tissue polarity phenotypes in these wing regions, clones in this region that are mutant for either fz or Vang show strong domineering nonautonomy (Vinson and Adler 1987; Figure 5, A–C). Even in these regions of mutant wings, however, it is more difficult to recognize domineering nonautonomy than in a wild-type wing, particularly if the effect is not dramatic. This is because the polarity is not as uniform as in a wild-type wing, and the presence of wild-type cells that produce multiple hairs cannot be used as evidence of domineering nonautonomy. Hence, in addition to scoring clones as either positive or negative for a domineering effect as we have in the past, we included the category of weak nonautonomy for clones where there was a hint of domineering nonautonomy, but where the effect was not convincing enough to score it as a clear positive. We scored 21 Vang A3 clones in fzR54/fzK21 mutant wings. None showed the strong domineering nonautonomy typical of Vang clones in an otherwise wild-type wing. Twelve of the 21 clones showed no domineering nonautonomy (Figure 6, A and B) while 9 were classified as producing weak domineering nonautonomy (Figure 6C). In those cases where we saw the weak nonautonomy, it was located proximal to the clone and hairs pointed away from the clone border as is seen for Vang clones in a wild-type background. Thus, we conclude that fz acts as a suppressor of the domineering nonautonomy of Vang.

View this table:
TABLE 3

Van Gogh function is required for a gradient of frizzled expression to produce a region of reversed wing hair polarity

Figure 5.

Domineering cell nonautonomy of Vang and fz shown in twin spots. A, B, and C show VangA3 kojVB13/VangA3 kojVB13 clones along with their pwn twin spots. The VangA3 kojVB13 clone is outlined in black and the pwn clone is shown in red. In A and B, note the wild-type cells proximal to the VangA3 kojVB13 clone that show abnormal polarity. The pwn twin spot, which should contain two wild-type copies of Vang, is not juxtaposed to the wild-type hairs showing altered polarity. D shows the equivalent experiment showing a fzR52 strb/fzR52 strb clone (outlined in black) and a flr/flr twin spot (outlined in red). Note the cells distal to the fzR52 strb clone with altered polarity and that the flr cells, which should have two doses of fz+, are not juxtaposed to this region.

We also carried out a complementary experiment in which we generated fz R52 strb clones in a Vang amorphic mutant background (Vang A3). Once again we only scored clones in regions where the polarity of hairs in a Vang wing was consistent. We scored 20 clones and found that 15 showed no domineering nonautonomy (Figure 7). The remaining 5 were scored as showing weak domineering nonautonomy. As is the case for fz clones in a wild-type wing, these clones affected the polarity of wild-type cells located distal, but not proximal, to the clone. Since more than 85% of fz clones typically show clear-cut domineering nonautonomy in a wild-type background (Vinson and Adler 1987; Jones and Adler 1996), Vang mutations act as suppressors of the domineering nonautonomy of fz.

DISCUSSION

Vang and the fz pathway: The cellular phenotype of Vang mutant wings differs from that seen previously for other tissue polarity genes. Most Vang alleles result in more multiple hair cells than are seen in a fz-like mutant and fewer than are seen in an in–like mutant. However, several results suggest that Vang is a member of the fz signaling/signal transduction pathway. The polarity phenotype of Vang mutants is quite similar to that seen in loss-of-function mutations in fz, dsh, in, fuzzy, mwh, and fritz. Thus, on the basis of its polarity phenotype it seems likely that Vang mutations inactivate the fz pathway. Further, the observation that dsh, in, and mwh are epistatic to Vang suggests that Vang functions upstream of these genes in the fz signaling/signal transduction pathway. Given that it has a phenotype in between that of the fz-like and in-like genes, a reasonable guess is that it functions between these two groups of genes. However, this is unlikely to be the case. The dsh gene, which appears to function between fz and in, has a fz-like mutant phenotype. dsh is required for the transduction of the fz signal (Krasnowet al. 1995) and acts cell autonomously (Klingensmithet al. 1994; Theisenet al. 1994), as do in (Parket al. 1996) and fuzzy (Collier and Gubb 1997). In contrast, we found that Vang is not required for the transduction of the fz signal and acts cell nonautonomously. Hence, Vang is unlikely to function downstream of dsh. It is possible that Vang acts downstream of fz (and upstream of dsh), but that it is not required for the transduction of the fz signal (e.g., it could be a negatively acting factor). However, as is discussed below the data do not suggest a simple quantitative relationship between fz and Vang. A second possibility is that Vang is upstream of both fz and dsh in the fz pathway. Alternatively, there could be two fz-dependent pathways (Vinson and Adler 1987; Parket al. 1994): a cell autonomous pathway, which includes genes such as dsh and in, and a cell nonautonomous pathway. The cell nonautonomous pathway would be logically upstream of the cell autonomous pathway and Vang could function there.

Figure 6.

A, B, and C show Vang A3 kojVB13/Vang A3 kojVB13 clones in the E region of fzR54/fzK21 wings. A similar region of a sibling fzR54/fzK21 wing without a clone is shown in D. In A and B the Vang A3 kojVB13/Vang A3 kojVB13 clones do not show any evidence of domineering nonautonomy. The clone in C shows an example of what we scored as weak domineering nonautonomy (see region with arrow). Note how this domineering nonautonomy is much less dramatic than is seen for Vang A3 kojVB13/Vang A3 kojVB13 clones in a wild-type wing (Figure 5).

The nonautonomy of Vang and fz: Clones mutant for Vang showed a remarkable anti-fz-like domineering nonautonomy, and this domineering nonautonomy was suppressed in a fz mutant background. It is important to note that wild-type hairs whose polarity was affected by a Vang clone pointed away from the clone. On the basis of previous data that showed that hairs point from cells of higher Fz levels toward cells of lower levels (Adleret al. 1997), it seemed possible that Vang mutant cells might have higher than normal fz activity (i.e., Vang is an inhibitor of fz activity). This could also explain the apparent enhancement by a Vang mutant background of the ability of late fz overexpression to induce an in-like phenotype (see Table 1). However, there are several observations that cannot be explained by this hypothesis. One is that the Vang polarity pattern resembles that seen with a loss of function of fz or other members of the fz signal transduction pathway such as dsh. The uniform overexpression of fz in the early pupae produces a tissue polarity phenotype, but not one with a polarity pattern whose details resemble the fz/in polarity pattern (Krasnow and Adler 1994). Thus, the polarity pattern seen in Vang mutant wings is not what is expected for an increase in fz activity. Further, the ability of Vang to dominantly enhance the phenotype of the weak fzR53 allele is not expected if Vang acts as a negative regulator of fz activity. Indeed, if this were the case we would predict that Vang mutations should suppress a weak fz allele such as fzR53. Similarly, the ability of fz loss-of-function mutations to enhance the Vang-dominant tissue polarity phenotype is the opposite of what we would expect if Vang acted as an inhibitor of fz activity. Further, if Vang mutations cause a tissue polarity phenotype by increasing fz activity, then we would expect that a fz gain of function would enhance the Vang mutant phenotype. While this may be the case when fz is overexpressed just before prehair initiation (as noted above), it was not observed with earlier overexpression (e.g., 12 hours before prehair initiation) or weak overexpression of fz (i.e., wild-type gene expression as well as the basal expression of a hs-fz gene). Finally, we found that fz domineering nonautonomy is suppressed in a Vang mutant background. If Vang mutations resulted in higher fz activity, than the difference in fz levels between the Vang;fz clone cells and their Vang;fz/+ neighbors would be enhanced compared to fz clones in a wild-type wing. This seems likely to enhance rather than suppress the domineering nonautonomy of fz. Thus, there are data that suggest a positive relationship between Vang and fz, as well as data that suggest an antagonistic relationship. We conclude that the relationship between Vang and fz is not simply quantitative.

Figure 7.

A, B, and C show fzR52 strb/fzR52 strb clones in the E region of Vang A3/VangA3 wings. Note the clones in A and B do not appear to alter the polarity of the neighboring cells. C shows an example of weak domineering nonautonomy (see region with arrow). Note how this domineering nonautonomy is much less dramatic than is seen for fzR52 strb/fzR52 strb clones in a wild-type wing (compare to Figure 4D). D shows the equivalent region of a sibling Vang A3/Vang A3 wing without a clone.

The extensive genetic interactions seen between fz and Vang, between pk and Vang, and between fz and pk (R. E. Krasnow and P. N. Adler, unpublished results) suggest that these genes are functionally quite close. However, as is discussed in depth earlier for Vang and fz, the interactions between Vang and pk and between fz and pk (R. E. Krasnow and P. N. Adler, unpublished results) also cannot be explained by a simple quantitative interaction. We suggest that the products of these genes may interact physically in a tissue polarity receptor complex, and that interactions at the protein level are responsible for the complex array of interactions detected in our genetic experiments. That the Vang protein might be part of a protein complex is also suggested by the observation that the phenotype of VangTBS42/VangTBS42 wings is stronger than that of VangTBS42/Df. This result argues that the antimorphic Vang protein produced by this allele is antagonizing the product of another gene, and this is likely to be because of the formation of an inactive protein complex.

Models for tissue polarity and Vang: Two types of models have been proposed to account for the role of fz in tissue polarity and its domineering cell nonautonomy (Adleret al. 1997). One type is typified by the cell-by-cell signaling model, which suggests that each cell becomes polarized because the Fz protein is activated unevenly across cells (Parket al. 1994; Adleret al. 1997). In one version of this model the binding of Wnt ligand to Fz protein on the proximal side of a cell was suggested to locally inactivate Fz (Figure 8A; Adleret al. 1997). Active Fz protein on the distal edge of the cell was suggested to activate several signal transduction pathways. One pathway would lead to prehair initiation at the distal edge of the cell. A second would lead to the release of Wnt ligand at the distal edge of the cell to polarize the next most distal cell. This released Wnt could be newly synthesized or could be molecules transported through the cell, as has been suggested for Wg (Hayset al. 1997). The domineering nonautonomy of fz would be caused by the failure of the clone cells to release signal. An alternative class of model is represented by the secondary signaling model, which suggests that the Fz protein is differentially activated in cells along the proximal/distal axis of the wing by the nonsaturating binding of a gradient morphogen (Zhenget al. 1995; Adleret al. 1997). Cells would then produce a secondary signal in amounts that were proportional to the fraction of Fz receptors that bound ligand (Figure 8C). Assessing the level of secondary signal produced by their neighbors would serve to polarize cells. These models differ fundamentally in that fz has both cell autonomous and cell nonautonomous functions in the cell-by-cell signaling model while it only has cell nonautonomous functions in the secondary signaling model; and the secondary signal model invokes a long range gradient of a diffusible polarity morphogen. As we have discussed elsewhere, both models have difficulty explaining some data, hence it is not clear which is closer to being correct (Adleret al. 1997). As is described below, it is possible to incorporate the data on Vang into both of these models.

In the cell-by-cell signaling model Vang can be placed into the signal transduction pathway (or transport pathway) that leads to the directional release of the Wnt ligand. To explain the domineering nonautonomy of Vang we need to hypothesize that Vang mutations result in signal being released in all directions, not just distally (Figure 8B). In this model Vang is involved in helping establish the spatial specificity of fz action and cannot be thought of as a simple positive or negative factor.

This is consistent with the data that do not show a simple quantitative relationship between fz and Vang. In the absence of Vang, fz signaling (i.e., release of ligand) would be equivalent in all directions, leading to a lack of polarized cells and a polarity phenotype that is similar to no fz signaling. This promiscuous signaling in a Vang mutant would be expected to suppress the consequences of the abnormal signaling caused by a fz clone, and hence suppress the domineering nonautonomy of fz. In the absence of fz function the domineering nonautonomy of Vang would be suppressed, as this nonautonomy is a consequence of abnormal fz signaling. Thus, the cell-by-cell signaling model can incorporate our observations on Vang.

Figure 8.

Shown are models to explain the roles of fz and Vang in wing tissue polarity. A shows a version of the cell-by-cell signaling model (Adleret al. 1997). Wnt ligand released at the distal edge of a cell binds to and inactivates Fz receptor on the proximal edge of neighboring cells. This leads to the activation of fz-dependent signal transduction pathways at the distal edge of cells. These signal transduction pathways lead to prehair initiation near the distal vertex resulting in hairs with distal polarity, to the release of Wnt ligand at the distal edge of the cell, and to the desensitization of Fz receptor in the distal part of the cell. The Vang gene is hypothesized to be part of the pathway that results in the distal release of Wnt. Hence, in a Vang mutant Wnt is released proximally as well as distally (B). C shows a version of the secondary signaling model. A gradient of a diffusible morphogen Wnt leads to a gradient of Fz receptor being activated by ligand binding. The activation of Fz receptor leads to the proportional production of a secondary signal. Cells are polarized because of a higher concentration of secondary signal on one side vs. the other. The Vang protein is hypothesized to be involved in coupling activated Fz to the production of secondary signal. Thus, in a Vang mutant a high constant level of secondary signal is produced (D).

In the secondary signaling model we can hypothesize that in a Vang mutant a similar high level of secondary signal is produced regardless of the degree of fz activation by ligand (Figure 8D). The uniform secondary signal would be equivalent to no secondary signal with respect to polarizing neighboring cells. In such a model Vang could be a negative regulator/modulator of fz signal transduction. Although it would be downstream of fz, as a negative regulator it would not be expected to be required for the transduction of the fz signal. This simple model has problems in explaining the interactions noted above that do not suggest a simple quantitative relationship between these two genes; but perhaps these could be explained as a consequence of direct interactions between the proteins.

It is possible that the connection between fz and Vang is not as close as is suggested above. For example, the activity of these genes could be separated in time. In the developing Drosophila eye it has been suggested that polar/equatorial polarity is established in two phases. The first signal comes from the pole and is dependent on wingless, while a later one is hypothesized to originate at the equator and is likely dependent on fz (Reifegersteet al. 1997; Wherli and Tomlinson 1998). If such a two-phase process functions in establishing wing tissue polarity it is possible that Vang could function in the early phase in the establishment of a polarizing signal that later functions in tissue polarity by polarizing fz activity. In a Vang mutant the polarizing signal could be ubiquitously present, leading to a failure to polarize fz activity. Vang clones could produce their proximal cell nonautonomy by causing an ectopic source of polarizing signal. The polarizing activity could represent the long-range morphogen hypothesized by the secondary signal model or an initiation center for cell-by-cell signaling.

The mutual suppression of the domineering nonautonomy of fz and Vang clones by mutations in Vang and fz, respectively, was not complete. The failure of Vang to completely suppress the domineering nonautomomy of fz could be because the Vang allele used in these experiments (VangA3) was not a null allele. The further characterization of Vang will be required to determine if this is a null allele. We also need to be able to explain the failure of fz mutations to completely suppress the domineering nonautonomy of Vang. The fz genotype used (fzR54/fzK21) in these experiments is expected to produce some protein (fzK21, which is a breakpoint in the first intron of fz, is a protein null, but fzR54 does produce Fz protein; Joneset al. 1996); however, this genotype is a phenotypic null with respect to wing hair polarity. Thus, we think that the failure to see complete suppression is not because of residual fz activity. It is possible that partial functional redundancy between fz and a second fz family member might be responsible for the residual Vang nonautonomy. The secondary signaling model can also be modified to account for the residual domineering nonautonomy without invoking any redundancy or mutations not completely inactivating genes. For example, some secondary signal could be produced in the absence of fz function and the amount of this signal could be modulated by Vang. Thus, clones of Vang cells in a fz wing could produce more secondary signal than their neighbors and hence still produce some domineering nonautonomy. Similarly, the level of secondary signal produced by clones of fz cells in an otherwise Vang wing might be less than their neighbors, leading to only partial suppression of the domineering nonautonomy of fz.

Acknowledgments

This work was supported by a grant from the National Institutes of Health (GM37136).

Footnotes

  • Note added in proof: We have determined that Van Gogh is allelic to strabismus (Wolff and Rubin, 1998. Development 125: 1149–1159).

  • Communicating editor: T. Schüpbach

LITERATURE CITED

View Abstract