ABSTRACT
The eye imaginal disc displays dorsal-ventral (D-V) and anterior-posterior polarity prior to the onset of differentiation, which initiates at the intersection of the D-V midline with the posterior margin. As the wave of differentiation progresses anteriorly, additional asymmetry develops as ommatidial clusters rotate coordinately in opposite directions in the dorsal and ventral halves of the disc; this forms a line of mirror-image symmetry, the equator, which coincides with the D-V midline of the disc. How D-V pattern is established and how it relates to ommatidial rotation are unknown. Here we address this question by assaying the expression of various asymmetric markers under conditions that lead to ectopic differentiation, such as removal of patched or wingless function. We find that D-V patterning develops gradually and that wingless plays an important role in setting up this pattern. We show that wingless is necessary and sufficient to induce dorsal expression of the gene mirror prior to the start of differentiation and also to restrict the expression of the WR122 marker to differentiating photoreceptors near the equator. In addition, we find that manipulations in wingless expression shift the D-V axis of the disc as evidenced by changes in the expression domains of asymmetric markers, the position of the site of initiation and the equator, and the pattern of epithelial growth. Thus, Wg appears to coordinately regulate multiple events related to D-V patterning in the developing retina.
INTRODUCTION
The secreted signaling molecules Hedgehog (Hh), Decapentaplegic (Dpp), Wingless (Wg), and their homologs play multiple roles during development of invertebrates and vertebrates (Blair, 1995; Burke and Basler, 1997; Roelink, 1996; Sasai and De Robertis, 1997; Serrano and O’Farrell, 1997). In the Drosophila eye, Hh, Dpp and Wg have been studied primarily for their effects on retinal differentiation in the eye imaginal disc of third instar larvae (for recent reviews see: Heberlein and Moses, 1995; Treisman and Heberlein, in press). Differentiation begins in the mid-third instar stage at the posteriormost tip of the disc, which coincides with its dorsal-ventral (D-V) midline. Differentiation progresses as a wave that sweeps across the eye imaginal disc in a posterior-to-anterior direction over the span of 2 days. The wave of differentiation is preceded by the morphogenetic furrow (MF), a depression in the retinal epithelium caused by local changes in cell shape (Ready et al., 1976; Tomlinson, 1985; Wolff and Ready, 1991). Cells located behind the MF differentiate in a well-characterized temporal sequence that is laid out spatially in a smooth gradient spanning the disc from the MF to the posterior margin (Tomlinson and Ready, 1987).
dpp is expressed along the posterior and lateral margins of the mid-third instar eye disc (Blackman et al., 1991; Masucci et al., 1990), where it promotes the initiation of differentiation (Chanut and Heberlein, 1997; Pignoni and Zipursky, 1997). In contrast, wg is expressed along the anterior lateral margins, where it acts as an inhibitor of differentiation (Ma and Moses, 1995; Treisman and Rubin, 1995). These antagonistic activities and complementary expression patterns are thought to restrict the onset of differentiation to a narrow region at the posterior margin (for recent review see Treisman and Heberlein, in press) and, later, to control the orderly progress of differentiation along the lateral margins (Chanut and Heberlein, 1997). Upon initiation of differentiation, hh is expressed in differentiating cells behind the MF (Lee et al., 1992; Ma et al., 1993); Hh function is necessary to ensure the anterior progression of the differentiation wave (Heberlein et al., 1993; Ma et al., 1993). While the exact mechanism underlying Hh function in the eye remains unresolved, activation of the Hh pathway induces the expression of several genes that play important roles in retinal development in cells located in and ahead of the MF (Heberlein et al., 1995; Pan and Rubin, 1995; Strutt et al., 1995).
Below the regular external array of the ∼800 facets that form the adult compound eye, individual ommatidia are polarized along the anterior/posterior (A-P) and D-V axes. Each eye contains ommatidia of two chiral forms that occupy the dorsal or ventral halves of the eye; these two types of ommatidia are separated by a line of mirror-image symmetry, the equator, which coincides with the D-V midline of the eye (Dietrich, 1909). Ommatidial polarization along the A-P axis arises as a consequence of the direction of MF progression as cells are recruited into specific positions along the A-P axis (Tomlinson, 1985; Tomlinson and Ready, 1987). As ommatidia continue to develop, they rotate by exactly 90°; those in the dorsal half of the eye rotate in the opposite direction from those in the ventral half. During this rotation, changes in cell contacts break the bilateral symmetry, generating the dorsal and ventral chiral forms. While the nature of the signal(s) that direct the choice of ommatidial rotation is unknown, it has been proposed that such a signal emanates from the equator and that it is mediated, directly or indirectly, by the product of the frizzled (fz) gene (Zheng et al., 1995). In addition to fz, mutations in spiny legs (sple), dishevelled (dsh) and RhoA disrupt the process by which ommatidia choose their direction of rotation (Gubb, 1993; Strutt et al., 1997). Mutations in nemo and roulette affect the execution of the rotation program at a later stage (Choi and Benzer, 1994).
Patterning of the disc along the A-P and D-V axes occurs prior to the start of differentiation as evidenced by the asymmetric expression patterns of several enhancer-trap lines (Brodsky and Steller, 1996). How this early patterning is established and to what extent it determines ommatidial polarization are unknown; investigations into a possible link between the two processes have led to contradictory conclusions. Ma and Moses postulated that ommatidial rotation is governed by a ‘long-range polarity field’ (Ma and Moses, 1995); in this model, each ommatidial cluster senses its position relative to the equator and the MF (the two-vector model) and rotates accordingly. By contrast, we (Chanut and Heberlein, 1995) and Strutt and Mlodzik (1995) found that ectopic ommatidial fields can generate their own equator; this led to the proposal that local cellular interactions are sufficient to generate an equator, independently of global polarity signals. Recently it has been shown that the homeobox-gene mirror, which is expressed in the dorsal eye disc prior to the start of differentiation, is required for proper equator formation (McNeill et al., 1997), supporting the role of early D-V patterning in ommatidial polarization and equator formation.
In this study, we explore further the causal relationship between D-V patterning and ommatidial rotation. Our results show that wg plays an important role in D-V patterning in the eye disc. We find that Wg regulates gene expression along the D-V axis, both ahead of and behind the MF; Wg induces the dorsal expression of mirror while repressing the expression of the equatorial marker WR122. We show that manipulations in wg expression can shift the D-V axis of the disc as monitored by the expression of asymmetric markers, the pattern of tissue growth, the location of the site of initiation of differentiation and ommatidial rotation. We discuss models for the coordinate regulation of D-V patterning and equator formation.
MATERIALS AND METHODS
Fly stocks
The following fly stocks were used: ptcB9 (Nüsslein-Volhard and Wieschaus, 1980), WR122-lacZ (enhancer trap insertion in 53C obtained from G. M. Rubin), wgIL114 (Nüsslein-Volhard and Wieschaus, 1980; obtained from J. E. Treisman), wg-lacZ (corresponds to wgen11 in Kassis et al., 1992); UAS-lacZ (line 4.2.4B; Brand and Perrimon, 1993), UAS-wg (Lawrence et al., 1995; obtained from S. M. Cohen), dpp-GAL4 (line 40C-6; Staehling-Hampton et al., 1994; this driver contains a region of the dpp enhancer, the blk domain, that fails to recapitulate the normal dpp expression pattern in the eye), dpp-lacZ (line H1-1; Blackman et al., 1991), Act5C>y+>wg (line 411.33; Struhl and Basler, 1993), mrr-lacZ (same as F7-lacZ, fails to complement mrr, line P69Df7 in Brodsky and Steller, 1996), fj-lacZ (enhancer-trap insertion in fj, line P55Ci20 in Brodsky and Steller, 1996), FL122 (Struhl and Basler, 1993), and FRT43D (Xu and Rubin, 1993). All other stocks used in this study were constructed using these reagents by standard genetic crosses.
ptc clones
y, FL122; FRT43D, ptc9B/CyO were crossed to y, w; FRT43D, arm-lacZ/CyO. y, FL122; FRT43D, ptc9B, WR122-lacZ/CyO were crossed to y, w; FRT43D, WR122-lacZ/SM6-TM6B. Larvae were grown at 25°C and subjected to a single 30 minute heat shock at 37°C between 24 and 48 (first instar) or 48 and 72 (second instar) hours of development and dissected as wandering third instar larvae.
wgts experiments
w; wgIL114, WR122-lacZ/SM6-TM6B larvae or the progeny of the cross of w; wgIL114; mrr-lacZ/SM6-TM6B and w; wgIL114/SM6-TM6B were grown at 17°C and shifted to 29°C at various times prior to dissection of Tb+ larvae.
Act5C>wg experiments
y, FL122; WR122-lacZ flies were crossed to y; Act5C>y+>wg flies, and progeny were grown at 17°C and heat shocked at 33.5°C for 20 or 30 minutes between 48 and 72 hours of development.
UAS-GAL4 experiments
The following crosses and stocks were used: (1) w; UAS-lacZ; dpp-GAL4/SM6-TM6B × w; UAS-wg (Fig. 3C); (2) w; dpp-lacZ; dpp-GAL4/SM6-TM6B × w; UAS-wg (Fig. 3E); (3) w; WR122-lacZ; dpp-GAL4/SM6-TM6B × w; WR122-lacZ; UAS-wg/SM6-TM6B (Fig. 3G); (4) w; mrr-lacZ, UAS-wg/SM6-TM6B × w; dpp-GAL4/TM6B (Figs 4E, 5A-D); (5) w; fj-lacZ; UAS-wg/SM6-TM6B × w; dpp-GAL4/SM6-TM6B (Fig. 5D); (6) w; UAS-lacZ; dpp-GAL4 × w; UAS-wg (Fig. 5G); (7) w; dpp-lacZ; UAS-wg/SM6-TM6B × w; dpp-GAL4/TM6B (Fig. 5H). Progeny were grown at 17°C and shifted to 25°C 24 or 48 hours prior to dissection of Tb+ larvae. If larvae were kept at 17°C and dissected as wandering third instar larvae, expression of the transgene had only minor consequences.
Asymmetric expression of the equatorial marker WR122-lacZ in ectopic ommatidia induced by local loss of patched function. ptc clones were induced in the second (A-J) or first (K,L) larval instar and analyzed in eye discs dissected from late third instar larvae. Discs were doubly stained with mAbBP104 or anti-Elav (brown) to identify differentiating photoreceptors, and with X-GAL (blue), to establish the pattern of β-galactosidase expression. Clones in A and B (white arrowheads) were identified by the loss of the arm-lacZ marker, clones in C-J (black arrowheads) by ectopic differentiation and the clone in K (black arrowhead) by the presence of abnormally structured and misrotated ommatidia. (A,B) A typical disc carrying several marked ptc clones displaying precocious differentiation that starts within the boundaries of the clone and spreads into the wild-type surrounding tissue. (C) Expression of the equatorial marker WR122-lacZ in wild type. (D-J) Expression of WR122-lacZ in ptc clones located in different positions in the disc. Expression of the marker is homogeneous in medial clones (D), asymmetric in lateral clones (E-H) or anterior marginal clones (J), and absent from lateral marginal clones (I). (K) A posterior ventral clone leads to the formation of an ectopic line of mirror-image symmetry (ventral dotted white line), which can be seen in the close-up (L) through the asymmetric expression of WR122-lacZ; the normal equator is recognized by the position of the optic stalk. Posterior is to the right, dorsal is up, and the position of the optic stalk is indicated in all figures with a horizontal line.
Asymmetric expression of the equatorial marker WR122-lacZ in ectopic ommatidia induced by local loss of patched function. ptc clones were induced in the second (A-J) or first (K,L) larval instar and analyzed in eye discs dissected from late third instar larvae. Discs were doubly stained with mAbBP104 or anti-Elav (brown) to identify differentiating photoreceptors, and with X-GAL (blue), to establish the pattern of β-galactosidase expression. Clones in A and B (white arrowheads) were identified by the loss of the arm-lacZ marker, clones in C-J (black arrowheads) by ectopic differentiation and the clone in K (black arrowhead) by the presence of abnormally structured and misrotated ommatidia. (A,B) A typical disc carrying several marked ptc clones displaying precocious differentiation that starts within the boundaries of the clone and spreads into the wild-type surrounding tissue. (C) Expression of the equatorial marker WR122-lacZ in wild type. (D-J) Expression of WR122-lacZ in ptc clones located in different positions in the disc. Expression of the marker is homogeneous in medial clones (D), asymmetric in lateral clones (E-H) or anterior marginal clones (J), and absent from lateral marginal clones (I). (K) A posterior ventral clone leads to the formation of an ectopic line of mirror-image symmetry (ventral dotted white line), which can be seen in the close-up (L) through the asymmetric expression of WR122-lacZ; the normal equator is recognized by the position of the optic stalk. Posterior is to the right, dorsal is up, and the position of the optic stalk is indicated in all figures with a horizontal line.
Ectopic expression of WR122-lacZ in discs from larvae carrying a temperature sensitive allele of wingless. Larvae homozygous for wgIL114 and WR122-lacZ were grown at the permissive temperature (17°C) and shifted to the restrictive temperature (29°C) at different times prior to dissection as late third instar larvae. Discs were doubly stained for the neuron-specific protein Elav (brown) and β-galactosidase activity (blue). (A) Wild-type control; wgIL114 larvae shifted to 29°C 24 hours prior to dissection. Limited ectopic differentiation is seen at the dorsal margin (arrowhead). (C) wgIL114 larvae shifted to 29°C 48 hours prior to dissection. Extensive ectopic differentiation is associated with ectopic expression of WR122-lacZ (between angled arrows). Lateral-posterior domains still lack WR122-lacZ expression (arrowheads).
Ectopic expression of WR122-lacZ in discs from larvae carrying a temperature sensitive allele of wingless. Larvae homozygous for wgIL114 and WR122-lacZ were grown at the permissive temperature (17°C) and shifted to the restrictive temperature (29°C) at different times prior to dissection as late third instar larvae. Discs were doubly stained for the neuron-specific protein Elav (brown) and β-galactosidase activity (blue). (A) Wild-type control; wgIL114 larvae shifted to 29°C 24 hours prior to dissection. Limited ectopic differentiation is seen at the dorsal margin (arrowhead). (C) wgIL114 larvae shifted to 29°C 48 hours prior to dissection. Extensive ectopic differentiation is associated with ectopic expression of WR122-lacZ (between angled arrows). Lateral-posterior domains still lack WR122-lacZ expression (arrowheads).
Inhibition of WR122-lacZ expression by ectopic expression of wingless. Eye discs from late third instar larvae were doubly stained with mAbBP104 (brown, A-G) or anti-Elav (brown; H,I) and X-GAL (blue). (A) Expression of wg-lacZ in wild type is restricted to the dorsal and ventral margins in the anterior eye disc. (B) dpp-GAL4 drives expression of UAS-lacZ along the dorsal and ventral margins ahead and behind the MF. (C) dpp-GAL4 drives the expression of UAS-wg and UAS-lacZ along the lateral margins. Differentiation is restricted to the median portion of the disc. (D) Expression of dpp-lacZ is restricted to the MF (white arrowhead) in wild type. (E) Expression of dpp-lacZ in dpp-GAL4/UAS-wg larvae reveals that the MF progresses as an arc, retarded along the margins. (F) Expression of WR122-lacZ in wild type. (G) Expression of WR122-lacZ in dpp-GAL4/UAS-wg is restricted to a narrow median region. (H,I) Expression of WR122-lacZ in discs carrying Act>wg clones (evidenced by the block in differentiation, arrowheads). Expression of the marker is lost near the clones (arrows in H,I). Because the disc in panel H is somewhat folded near the clone (due to local overgrowth), expression of WR122-lacZ appears to reach the anterior edge of differentiation.
Inhibition of WR122-lacZ expression by ectopic expression of wingless. Eye discs from late third instar larvae were doubly stained with mAbBP104 (brown, A-G) or anti-Elav (brown; H,I) and X-GAL (blue). (A) Expression of wg-lacZ in wild type is restricted to the dorsal and ventral margins in the anterior eye disc. (B) dpp-GAL4 drives expression of UAS-lacZ along the dorsal and ventral margins ahead and behind the MF. (C) dpp-GAL4 drives the expression of UAS-wg and UAS-lacZ along the lateral margins. Differentiation is restricted to the median portion of the disc. (D) Expression of dpp-lacZ is restricted to the MF (white arrowhead) in wild type. (E) Expression of dpp-lacZ in dpp-GAL4/UAS-wg larvae reveals that the MF progresses as an arc, retarded along the margins. (F) Expression of WR122-lacZ in wild type. (G) Expression of WR122-lacZ in dpp-GAL4/UAS-wg is restricted to a narrow median region. (H,I) Expression of WR122-lacZ in discs carrying Act>wg clones (evidenced by the block in differentiation, arrowheads). Expression of the marker is lost near the clones (arrows in H,I). Because the disc in panel H is somewhat folded near the clone (due to local overgrowth), expression of WR122-lacZ appears to reach the anterior edge of differentiation.
wingless regulates mirror expression. Eye discs from mid-third or late-third (B-D) instar larvae were doubly labeled with mAbBP104 (brown) and X-GAL (blue). (A,B) Expression of an enhancer-trap insertion in mirror (mrr-lacZ) in wild-type discs is restricted to the dorsal disc ahead of the MF. (C) Expression of mrr-lacZ in discs from wgIL114 larvae shifted from the permissive to the restrictive temperature 24 hours prior to dissection. Expression of mrr-lacZ is reduced in the retina. Expression in the antennal disc appears less sensitive to the reduction of wg function. Ectopic differentiation induced by overexpression of dpp does not lead to a loss of mrr-lacZ expression (data not shown). (D) Expression of mrr-lacZ in dpp-GAL4/UAS-wg larvae is expanded ventrally. The Bolwig nerve (arrowhead in all panels) marks the position of the original D-V midline.
wingless regulates mirror expression. Eye discs from mid-third or late-third (B-D) instar larvae were doubly labeled with mAbBP104 (brown) and X-GAL (blue). (A,B) Expression of an enhancer-trap insertion in mirror (mrr-lacZ) in wild-type discs is restricted to the dorsal disc ahead of the MF. (C) Expression of mrr-lacZ in discs from wgIL114 larvae shifted from the permissive to the restrictive temperature 24 hours prior to dissection. Expression of mrr-lacZ is reduced in the retina. Expression in the antennal disc appears less sensitive to the reduction of wg function. Ectopic differentiation induced by overexpression of dpp does not lead to a loss of mrr-lacZ expression (data not shown). (D) Expression of mrr-lacZ in dpp-GAL4/UAS-wg larvae is expanded ventrally. The Bolwig nerve (arrowhead in all panels) marks the position of the original D-V midline.
Ectopic wingless shifts the D-V axis of the disc. Eye-antennal disc complexes from mid to late third instar larvae were doubly labeled with mAbBP104 (brown) and X-GAL (blue). In all discs, the Bolwig nerve (arrowhead) indicates the position of the original midline of the disc. (A-D) Expression of mrr-lacZ in dpp-GAL4/UAS-wg larvae (for wild-type controls see Fig. 4A,B). In discs where differentiation is blocked (or delayed), the ventral border of mrr-lacZ expression is shifted clockwise by approximately 45° (A). An extreme example, where this shift is approximately 90°, is shown in B. Differentiation, when present, always starts at the ventral-posterior border of the mrr-lacZ expression domain (arrow in C and D) and progresses as an arc in an anterior-dorsal direction (D). (E,F) Expression of fj-lacZ in eye discs from wild-type (E) and dpp-GAL4/UAS-wg larvae (F). fj-lacZ is normally expressed in a pattern centered at the D-V midline ahead of the MF and in a region near the developing ocelli (white arrowhead) (E). In a dpp-GAL4/UAS-wg disc (F), in which the start of differentiation is shifted ventrally (black arrow), reduced fj-lacZ expression is observed in an oblique pattern corresponding with the shifted D-V midline (white arrow in F). (G) The pattern of ectopic wg expression in dpp-GAL4/UAS-wg eye discs was monitored by the expression of UAS-lacZ. The point of lowest expression corresponds to the point that defines the new midline (arrow). (F) Expression of dpp-lacZ in the eye disc of dpp-GAL4/UAS-wg larvae is restricted to the posterior-ventral margin, which coincides with the region of lowest wg expression. While the expression of the dpp-lacZ reporter is changed by ectopic Wg, expression of the dpp-GAL4 driver, which contains a smaller region of the dpp enhancer (see Materials and Methods), appears unaffected.
Ectopic wingless shifts the D-V axis of the disc. Eye-antennal disc complexes from mid to late third instar larvae were doubly labeled with mAbBP104 (brown) and X-GAL (blue). In all discs, the Bolwig nerve (arrowhead) indicates the position of the original midline of the disc. (A-D) Expression of mrr-lacZ in dpp-GAL4/UAS-wg larvae (for wild-type controls see Fig. 4A,B). In discs where differentiation is blocked (or delayed), the ventral border of mrr-lacZ expression is shifted clockwise by approximately 45° (A). An extreme example, where this shift is approximately 90°, is shown in B. Differentiation, when present, always starts at the ventral-posterior border of the mrr-lacZ expression domain (arrow in C and D) and progresses as an arc in an anterior-dorsal direction (D). (E,F) Expression of fj-lacZ in eye discs from wild-type (E) and dpp-GAL4/UAS-wg larvae (F). fj-lacZ is normally expressed in a pattern centered at the D-V midline ahead of the MF and in a region near the developing ocelli (white arrowhead) (E). In a dpp-GAL4/UAS-wg disc (F), in which the start of differentiation is shifted ventrally (black arrow), reduced fj-lacZ expression is observed in an oblique pattern corresponding with the shifted D-V midline (white arrow in F). (G) The pattern of ectopic wg expression in dpp-GAL4/UAS-wg eye discs was monitored by the expression of UAS-lacZ. The point of lowest expression corresponds to the point that defines the new midline (arrow). (F) Expression of dpp-lacZ in the eye disc of dpp-GAL4/UAS-wg larvae is restricted to the posterior-ventral margin, which coincides with the region of lowest wg expression. While the expression of the dpp-lacZ reporter is changed by ectopic Wg, expression of the dpp-GAL4 driver, which contains a smaller region of the dpp enhancer (see Materials and Methods), appears unaffected.
Immunohistochemistry
Third instar eye imaginal discs were immunostained with the following antibodies: rat-anti-Elav (a gift from G. M. Rubin), mAbBP104 (a gift from C. S. Goodman), or rabbit anti-Bar (a gift from K. Saigo) essentially as described before (Kimmel et al., 1990). Secondary antibodies used were goat anti-mouse-HRP (Biorad), donkey anti-rat-HRP, donkey anti-rabbit-FITC, or donkey anti-mouse-Cy5 (Jackson Laboratories). Discs were viewed and photographed with a Leica DMR microscope (Figs 1-5) or a Biorad MRC 1024 confocal microscope.
RESULTS
Ectopic ommatidial fields reveal a dorsoventral pattern ahead of the morphogenetic furrow
To determine if positional information associated with equator formation is present along the D-V axis of the disc ahead of the MF, we studied the expression of an equatorial marker in various genetic conditions that lead to ectopic neuronal differentiation. This marker, WR122, is a lacZ insertion in an unknown locus located at cytological position 53C (hereafter called WR122-lacZ). WR122-lacZ is expressed behind the MF in a sector centered about the equator (Fig. 1C). This expression is dependent on the activity of the gene fz, which is required for proper ommatidial rotation (Zheng et al., 1995). Thus, the patterned expression of WR122-lacZ coincides temporally and spatially with the process of equator formation and is, at least in part, under the same genetic control.
We first assayed the expression of the marker in small fields of ectopic ommatidia in the anterior eye disc induced by local removal of ptc function. As previously reported (Chanut and Heberlein, 1995; Ma and Moses, 1995; Strutt and Mlodzik, 1995; Wehrli and Tomlinson, 1995), loss of ptc function ahead of the MF activates the Hh pathway. This leads to precocious neuronal differentiation that spreads into the surrounding wild-type tissue. ptc clones were induced during the second larval instar and the consequences assayed in eye discs dissected from late third instar larvae. In contrast to clones induced in first instar larvae (Chanut and Heberlein, 1995), those induced in the second instar are relatively small and have only minor effects on disc growth and morphology (Fig. 1A,B).
The pattern of WR122-lacZ expression in ectopic ommatidial groups varied with the position of individual clones within the epithelium. Differentiating photoreceptors were identified by the expression of a neuronal form of neuroglian recognized by mAbBP104 (Hortsch et al., 1990). Internal ectopic fields located near the disc’s D-V midline always expressed WR122-lacZ homogeneously (Fig. 1D). In contrast, expression of the marker was restricted to a subset of ommatidia, always to those closest to the D-V midline, in clones located more laterally (Fig. 1E-H). We do not believe that WR122-lacZ expression in these lateral fields is restricted to homozygous mutant ptc tissue: while the mutant tissue is always located at the center of ectopic fields (Fig. 1A,B), WR122-lacZ expression can be associated with immature clusters that usually form in wild-type tissue surrounding the clone (Fig. 1E-H).
Ectopic ommatidia that arise in clones located near the disc’s margin also displayed differential WR122-lacZ expression depending on their position (Fig. 1I,J). WR122-lacZ was not expressed in ectopic fields arising at or very near the ventral or dorsal margins (Fig. 1I, data not shown). However, within a field that stretches along the ventral-anterior margin (Fig. 1J), the marker was expressed in a subset of ommatidia, those closest to the disc’s D-V midline. Thus, the potential to express WR122-lacZ is restricted to neurons located near the D-V midline regardless of their position along the A-P axis of the disc. This suggests that, although WR122-lacZ is normally expressed only behind the MF, the information necessary to restrict its expression exists ahead of the MF. Curiously, internal lateral ommatidial fields express WR122-lacZ asymmetrically regardless of their position in the disc (compare clones in Fig. 1E and G). This suggests that ommatidia in these fields respond to relative rather than absolute levels of a signal that exists along the D-V axis (see Discussion). Because the direction of ommatidial rotation cannot be ascertained in small ommatidial fields, it remains unclear whether the ability of clones to express WR122-lacZ is correlated with ommatidial polarization.
Different results are obtained if ptc clones (Fig. 1K,L), or clones that express hh ectopically (Chanut and Heberlein, 1995), are induced earlier in development, in the first larval instar. The resulting ectopic ommatidial fields are larger, they often fuse with the normal field and they arise most commonly near the margin. In these discs, WR122-lacZ is nearly always expressed, in addition to its normal domain, in a restricted pattern that centers approximately along the midline of the ectopic ommatidial field. An example of a ptc clone that led to ectopic differentiation along the posterior ventral epithelium is shown in Fig. 1K. In addition to the normal domain of WR122-lacZ expression aligned with the optic stalk, a second distinct sector of marker expression is observed more ventrally. Although the homozygous mutant ptc tissue could not be marked in this experiment, its position can be inferred from the presence of abnormally constructed and misrotated ommatidia at the posterior ventral margin (Fig. 1K, arrowhead). Closer examination of this disc shows that, in the ventral WR122-lacZ sector, a line can be drawn to separate ommatidia rotating in opposite orientations, indicating the formation of a line of mirror-image similar to an equator (Fig. 1K,L). Therefore, ectopic WR122-lacZ expression coincides, as in a normal disc, with the region where an equator forms, suggesting that expression of the marker is associated with ommatidial polarization.
One possible explanation for the different results obtained for WR122-lacZ expression in ectopic ommatidial fields induced at different times during retinal morphogenesis is that positional information along the D-V axis develops gradually. Thus, clones induced late in development reveal differences in positional information that exist along the D-V axis, while clones induced earlier may have the ability to set up this information, locally and independently from the disc’s normal D-V midline. Another possibility is that WR122-lacZ is expressed only during ommatidial polarization and equator formation and that this ability varies in different regions of the eye disc. In an attempt to distinguish these two possibilities, we assayed the expression of WR122-lacZ under conditions that dissociate ectopic ommatidial differentiation and equator formation, specifically, in conditions of reduced Wg activity.
Wg patterns the expression of an equatorial marker
Wg is necessary to restrict WR122-lacZ expression
A reduction in Wg function during the late larval stages promotes precocious differentiation in the eye disc. This differentiation starts from the dorsal (and to a lesser degree the ventral) margin and proceeds inwards, roughly perpendicularly to the direction of progression of the normal differentiation front. The ectopic ommatidia do not rotate and, as a consequence, an ectopic equator fails to form among them (Ma and Moses, 1995).
We determined the pattern of WR122-lacZ expression in these ectopic ommatidial fields in larvae carrying both the WR122-lacZ marker and a temperature-sensitive allele of wg, wgIL114 (Nüsslein-Volhard and Wieschaus, 1980; see Materials and Methods). Photoreceptor differentiation was monitored by the expression of the neuron-specific protein Elav (Robinow and White, 1991). The origin of ommatidia, whether ectopic or normal, can be ascertained based on their location, maturity and orientation relative to the A-P and D-V axes of the disc (Ma and Moses, 1995; Treisman and Rubin, 1995). Larvae were shifted from the permissive (17°C) to the restrictive temperature (29°C) at various times during development and dissected in the late third instar stage. Eye discs from larvae that were shifted 24 hours prior to dissection displayed only limited ectopic differentiation along the dorsal margin (Fig. 2B, arrowhead), and expression of WR122-lacZ was relatively normal. Larvae grown at the restrictive temperature for 48 hours showed extensive precocious differentiation and ectopic WR122-lacZ expression (Fig. 2C). Thus, in contrast to the limited expression of WR122-lacZ in ectopic ommatidia induced by ptc clones (Fig. 1), expression of the marker is unrestricted among ectopic ommatidia that differentiate as a consequence of reduced Wg function (Fig. 2C between arrows). In addition, the normal expression domain of the marker is broadened towards the lateral margins (most obvious near the ventral margin of the disc shown in Fig. 2C).
We conclude that expression of WR122-lacZ per se is not a direct consequence of ommatidial polarization. Rather, expression of the marker is inhibited by Wg in ommatidia located near the disc’s margin, which restricts expression to the equatorial region. Consistent with this role, wg is expressed along the lateral margins of the disc ahead of the MF (Fig. 3A; Ma and Moses, 1995; Treisman and Rubin, 1995). The domain in which WR122-lacZ is normally expressed (Figs 1C, 2A) occurs a significant distance away from the region of wg transcription (Fig. 3A), implying that wg may act at a distance to inhibit the marker’s expression. Whether this is achieved directly, by diffusion of Wg protein, or indirectly, through another diffusible molecule, is not known. Nevertheless, Wg has been shown to act as a long-range morphogen in other imaginal discs (Neumann and Cohen, 1997; Zecca et al., 1996).
Wg is sufficient to inhibit WR122-lacZ expression
Our experiments showed that wg is necessary for the restriction of WR122-lacZ expression to the equatorial region of the eye disc. To test if wg is sufficient to inhibit the marker’s expression and to establish spatial requirements for such an inhibition, we determined the effects of ectopic Wg on WR122-lacZ expression. We first used the UAS/GAL4 system (Brand and Perrimon, 1993) to drive wg expression along the lateral margins of the disc both ahead of and behind the MF. Specifically, a dpp-GAL4 line (Staehling-Hampton et al., 1994; see Materials and Methods) was used to drive UAS-wg (Lawrence et al., 1995) in the pattern shown in Fig. 3B. The levels of ectopic Wg were modulated taking advantage of the temperature sensitivity of the UAS/GAL4 system (Brand et al., 1994; see Materials and Methods). Larvae were raised at 17°C to ensure low levels of ectopic Wg during the early stages of development. To increase GAL4 activity and ectopic wg expression, larvae were shifted to 25°C 24 hours prior to dissection. Expression of wg in this pattern partially blocks differentiation near the margins, as previously reported for the UAS-wgts construct (Treisman and Rubin, 1995; Fig. 3C). The curved expression of a dpp-lacZ reporter leading the front of differentiation (compare Fig. 3D and E) suggests that MF progression is merely delayed, as a complete block in MF progression leads to loss of dpp-lacZ expression (Heberlein et al., 1993).
Expression of WR122-lacZ in the UAS-wg/dpp-GAL4 background, while normal regarding its general location, is reduced to a narrow sector (compare Fig. 3F and G). We conclude that providing excess Wg at the dorsal and ventral margins is sufficient to repress WR122-lacZ expression in the more central portions of the retinal epithelium.
To test the effect of ectopic Wg inside the retinal epithelium, we generated clones of cells expressing wg under the control of the Actin 5C promoter (Act>wg clones, see Materials and Methods). Such clones, generated by the ‘flp-out’ method (Struhl and Basler, 1993), have been shown to inhibit retinal differentiation and alter the polarity of adjacent ommatidia (Treisman and Rubin, 1995; Tomlinson et al., 1997). While the exact position of these Act>wg clones could not be identified in these experiments, their presence can be inferred from the associated lack of differentiation (Fig. 3H,I; arrowheads). An Act>wg clone located near the center of the disc (Fig. 3H) arrested the anterior progress of differentiation and repressed WR122-lacZ expression in ommatidia located for the most part within 3 rows (Fig. 3H, arrow; see figure legend). A more dramatic effect can be observed in the disc shown in Fig. 3I, in which an Act>wg clone arose near the posterior D-V midline (position of the optic stalk is indicated by a horizontal line); expression of WR122-lacZ is shifted dorsally and eliminated from a domain along the clone estimated to be six ommatidia wide.
These observations show that ectopic wg expression can inhibit the expression of WR122-lacZ regardless of the location of Wg-expressing cells within the disc. Although the exact borders of the unmarked Act>wg clones are not known, this experiment shows that WR122-lacZ expression is inhibited by Wg over greater distances than ommatidial differentiation and suggests a difference in the threshold for each effect.
Wg controls expression of the dorsal gene mirror
The experiments described above show that wg can pattern the expression of an equatorial marker ahead of and behind the MF. mirror, a homeobox gene required for proper equator formation (McNeill et al., 1997), is expressed in the dorsal half of the eye ahead of the MF (Brodsky and Steller, 1996; Fig. 4A,B). To determine if wg is involved in setting up the pattern of mirror expression, we established the effect of both loss of wg function or ectopic wg expression on the expression of an enhancer-trap insertion in the gene (hereafter called mrr-lacZ; Brodsky and Steller, 1996). The ventral margin of mrr-lacZ expression coincides closely with the position of the Bolwig nerve (Fig. 4A,B, arrowheads), which serves as a good marker for the D-V midline.
To test the effect of loss of wg function on mrr-lacZ expression, larvae carrying both the wgIL114 allele and mrr-lacZ were shifted to the restrictive temperature and dissected 24 hours later as late third instar larvae. Expression of mrr-lacZ in the eye disc is dramatically reduced in these discs (Fig. 4C). Thus, high levels of wg activity are required to maintain expression of mrr-lacZ in the dorsal eye disc.
To address whether wg is sufficient to induce ventral mrr-lacZ expression, we used the UAS/GAL4 system as described above (Fig. 3). Larvae carrying UAS-wg, dpp-GAL4 and mrr-lacZ were grown at 17°C and shifted to 25°C 48 hours prior to dissection and histochemical analysis (see Materials and Methods). The phenotypes observed depended on the exact time at which larvae were shifted to the higher temperature (see below). If larvae were shifted during the early third instar, differentiation was blocked completely by overexpression of wg along the margins and mrr-lacZ expression was expanded ventrally (Fig. 4D). In some instances, expression of mrr-lacZ covered the entire disc (data not shown). Thus, wg appears necessary to maintain normal mrr-lacZ expression and sufficient to induce inappropriate mrr-lacZ expression in the ventral eye disc. It is possible that, during normal development, Wg induces a gradient of mrr transcription; autoregulation, a common feature of HD proteins, would later generate the sharp ventral border of mrr expression (McNeill et al., 1997). We showed previously that ectopic activation of the Hh pathway can lead to ectopic expression of mrr-lacZ in the ventral retina (Chanut and Heberlein, 1995). We now believe that this induction is mediated by Wg, which we have found to be induced along the anterior margin of the eye disc upon ectopic hh expression (data not shown).
Ectopic wg expression can generate a new D-V axis
In most larvae carrying dpp-GAL4 and UAS-wg that were shifted to 25°C 48 hours prior to dissection as third instar larvae the ventral margin of mrr-lacZ is rotated clockwise by approximately 45° relative to the original D-V midline of the disc, as visualized by the position of the Bolwig nerve (Fig. 5A). In some cases, this angle can reach nearly 90° (Fig. 5B). Differentiation, when present, always begins ventrally, and at the exact border between cells that express mrr-lacZ and those that do not (Fig. 5C,D; arrows). Occasionally, differentiation can progress part way across the disc, although with a curved front (Fig. 5D); an equator forms in these fields along the new midline (see Fig. 6 and below). We believe that these phenotypes reflect a shift in the location of the D-V midline rather than a change in position of the Bolwig nerve and optic stalk (used as a reference for the position of the original midline); both these structures are formed normally along the disc’s midline in dpp-GAL4/UAS-wg larvae prior to the temperature shift that increases ectopic Wg expression (data not shown) and it is unlikely that, once formed, these structures change their position.
Ectopic wingless shifts the position of the equator. Eye discs from wandering third instar larvae were doubly labeled with anti-Bar (green) and mAbBP104 (red). High levels of Bar expression labels the nuclei of R1 and R6, which mark the equational side of each ommatidial cluster. The dot of mAbBP104 staining marks the approximate center of each cluster and serves as a reference for the position of R1 and R6. (A,B) Disc from a wild-type larva. The equator (arrow) coincides with optic stalk (arrowhead). (B) A close-up clearly shows the equator (dotted line). (C,D) Disc from a dpp-Gal4/UAS-wg larva shifted from 17 to 25°C 48 hours prior to dissection. The front of differentiation is curved. An equator (white arrow) is located ventrally to the position of the original D-V midline marked by the position of the optic stalk (arrowhead). (D) A close-up clearly shows the presence of an equator (dotted line).
Ectopic wingless shifts the position of the equator. Eye discs from wandering third instar larvae were doubly labeled with anti-Bar (green) and mAbBP104 (red). High levels of Bar expression labels the nuclei of R1 and R6, which mark the equational side of each ommatidial cluster. The dot of mAbBP104 staining marks the approximate center of each cluster and serves as a reference for the position of R1 and R6. (A,B) Disc from a wild-type larva. The equator (arrow) coincides with optic stalk (arrowhead). (B) A close-up clearly shows the equator (dotted line). (C,D) Disc from a dpp-Gal4/UAS-wg larva shifted from 17 to 25°C 48 hours prior to dissection. The front of differentiation is curved. An equator (white arrow) is located ventrally to the position of the original D-V midline marked by the position of the optic stalk (arrowhead). (D) A close-up clearly shows the presence of an equator (dotted line).
The expression of an enhancer-trap insertion in fourjointed (fj-lacZ, Brodsky and Steller, 1996), which normally centers about the D-V midline ahead of the MF (Fig. 5E), is also shifted clockwise to reflect the new D-V midline (Fig. 5F). In addition, expression of fj-lacZ is strongly reduced upon ectopic wg expression (Fig. 5F and data not shown), suggesting that Wg normally restricts the expression of fj to the D-V midline.
The shifted site of initiation coincides with the location of lowest ectopic wg expression as revealed by the expression of the UAS-lacZ reporter in the dpp-GAL4/UAS-wg background (Fig. 5G). This in turn coincides with the point of highest residual dpp expression (Fig. 5H; see figure legend). In addition to the ventral shift in the site of initiation of differentiation, other manifestations of D-V patterning are also affected. First, growth of dorsal tissue (located above the Bolwig nerve, arrowheads) is reduced compared to the ventral half. Second, an oblique groove in the disc, probably reflecting local overgrowth, is aligned with the new border of mrr-lacZ expression (Fig. 5B-D). This groove, also visible in Fig. 5G and H, clearly bisects the disc into two symmetrical domains and appears to demarcate the new D-V midline.
To determine if the direction of ommatidial rotation and, consequently, the position of the equator was altered to reflect the disc’s new D-V midline, we ascertained ommatidial rotation in discs dissected from dpp-GAL4/UAS-wg larvae shifted from 18 to 25°C 48 hours prior to their dissection as wandering third instar larvae (Fig. 6). Ommatidial rotation was monitored using the anti-Bar antibody, which labels the nuclei of differentiating R1 and R6 cells (Higashijima et al., 1992); these cells identify the equatorial side of each ommatidial cluster (Fig. 6B). As shown above (Fig. 5D), the front of differentiation is curved in the dpp-GAL4/UAS-wg discs (Fig. 6C). An equator is clearly present in these discs (arrow in Fig. 6C and dotted line in Fig. 6D); this equator, however, is not aligned with the optic stalk (arrowhead in Fig. 6C), which marks the original D-V midline of the disc. Thus, ectopic expression of wg shifts the position of the equator ventrally, a location that is likely to coincide with the new D-V midline of these discs.
These data can be summarized as follows. First, overexpression of wg in an apparently symmetrical pattern along the dorsal and ventral margins acts asymmetrically, shifting the D-V axis of the disc and the position of the equator. Second, expression of high levels of wg along the disc’s margin inhibits the expression of dpp; while dpp is normally expressed all along the posterior and lateral margins of the disc prior to differentiation (Blackman et al., 1991; Masucci et al., 1990), only a small region of dpp expression remains near the domain of lowest wg expression. Because dpp is required for its own expression along the disc’s margins (Chanut and Heberlein, 1997), it is possible that Wg interferes with Dpp function, not its expression (Treisman and Rubin, 1995). Third, the site of initiation of differentiation coincides exactly with the point of intersection of the new midline with the posterior margin; dpp, which is required for the initiation of differentiation (Wiersdorff et al., 1996; Chanut and Heberlein, 1997), is expressed in a domain that includes the point of initiation. Finally, growth of the disc is also coordinated by this new D-V axis as it occurs symmetrically on both sides of the new midline.
DISCUSSION
The developing eye disc exhibits obvious signs of polarization as ommatidial assembly and rotation occur asymmetrically with respect to the epithelium’s A-P and D-V axes. An issue unresolved is whether ommatidia respond to global or local signals (or both) to direct their rotation and chirality; studies addressing this question have resulted in opposing conclusions (see Introduction). The study reported here was designed to clarify the nature and position of signals involved in polarization of the eye disc. Our results reveal a novel role for wg in patterning the disc along the D-V axis. This, together with the ability of wg to affect ommatidial chirality (Treisman and Rubin, 1995; Tomlinson et al., 1997), suggests that wg coordinates both D-V patterning and ommatidial polarization.
Where does wg act?
While the expression pattern of wg is fairly symmetrical (although the levels are higher dorsally, Fig. 3A), loss of wg function leads to both symmetric and asymmetric effects (Fig. 7A,B): wg normally induces the expression of mrr only in the dorsal retina and represses the expression of WR122-lacZ near the dorsal and ventral margins. Similarly, ectopic expression of wg (using the dpp-GAL4 driver) in an apparently symmetrical pattern (Fig. 3B,C and antibody staining, data not shown) leads to symmetric and asymmetric consequences: the site of initiation of differentiation, when altered, is always shifted ventrally (Fig. 5), while the expression of WR122-lacZ is inhibited equally in the vicinity of both margins (Fig. 3G). Thus, the ability of wg to pattern the disc along the D-V axis is regulated by additional factors that act selectively in the dorsal or ventral disc. For example, a repressor of mrr expression is present ventrally (Fig. 7B, factor X); the function of this repressor(s), can be overcome, however, by high levels of Wg in the ventral disc early during development (Fig. 4D). Genes of the Polycomb group (PcG) are good candidates for this repressor(s) as mrr expression is derepressed ventrally in various PcG mutants (D. Coen, personal communication).
Possible models for the role of wingless in dorsoventral patterning. (A) Cartoon summarizing the domains of expression of the various genes and markers used in this study. (B) Summary of the genetic interactions between wg and various genes and markers shown in A. Arrows represent positive interactions, while intersecting lines show negative interactions. (C,D) Nonmutually exclusive models for the role of wg in dorsoventral patterning. (C) The primary role of wg would be to establish the D-V midline; the midline in turn would help establish and/or maintain the equator. (D) wg would play an independent role in both the establishment of the D-V midline and in directing ommatidial rotation. See text for details.
Possible models for the role of wingless in dorsoventral patterning. (A) Cartoon summarizing the domains of expression of the various genes and markers used in this study. (B) Summary of the genetic interactions between wg and various genes and markers shown in A. Arrows represent positive interactions, while intersecting lines show negative interactions. (C,D) Nonmutually exclusive models for the role of wg in dorsoventral patterning. (C) The primary role of wg would be to establish the D-V midline; the midline in turn would help establish and/or maintain the equator. (D) wg would play an independent role in both the establishment of the D-V midline and in directing ommatidial rotation. See text for details.
While wg is expressed primarily ahead of the MF, where it patterns the disc along the D-V axis, wg also affects cells located behind the MF, as expression of WR122-lacZ is derepressed near the dorsal and ventral margins upon reduction of Wg function (Fig. 2). Whether the expression of D-V asymmetric markers is regulated directly by a gradient of wg activity or indirectly by a wg-induced secondary signal is unknown. However, WR122-lacZ expression does not appear to be a direct molecular read-out of absolute Wg levels. First, lateral ptc clones express the marker asymmetrically regardless of their proximity to the margin and the source of Wg (Fig. 1E,G). Second, ectopic expression of wg within the disc or along its margins appears to shift rather than completely erase expression of WR122-lacZ (Fig. 3I). Taken together, these observations suggest that WR122-lacZ expression is restricted to ommatidia located within the domain of lowest Wg function, even if this ‘relatively’ low level of Wg is higher than that normally sufficient to inhibit the marker’s expression in ommatidia located near the margins.
When does wg act?
In this study, our manipulations of Wg function and expression relied on the use of temperature-sensitive systems. Because the rate of larval development is altered by reduced and increased Wg function and because the kinetics of Wg inactivation (in wgts) or overexpression (in UAS/GAL4) are unknown, we could not infer the precise stage of discs at the time of the temperature shifts. However, relative times of Wg’s actions could be inferred from the phenotypes observed after various fixed times at particular temperatures. For example, early overexpression of wg was required for complete derepression of ventral mrr-lacZ expression (Fig. 4D); overexpression started at later times caused a clockwise rotation in the pattern of mrr-lacZ expression (Fig. 5A-D, F-H), while even later shifts resulted only in delayed MF progression near the margins (Fig. 3C,E) without affecting mrr-lacZ expression (data not shown). These observations suggest that the developing eye disc becomes gradually more resistant to Wg-induced manipulations of D-V patterning.
The effects of changes in wg function on WR122-lacZ expression suggest that wg acts relatively late to restrict expression to the equatorial region: WR122-lacZ expression is already slightly broadened in wgts larvae grown for only 24 hours at the restrictive temperature. This, together with the finding that wg patterns mrr expression, suggests that wg is required throughout retinal development for proper D-V patterning. In addition, we show that, as previously proposed (Heberlein and Moses, 1995) the site of initiation of differentiation along the disc’s margin is determined by the point of lowest wg expression (Fig. 5). Therefore, wg also plays a role, albeit indirect, in A-P patterning by helping define the site of initiation.
How does D-V midline establishment relate to equator formation?
While the D-V midline and the equator coincide spatially, they can be dissociated by various genetic manipulations. For example, mutations in sple (Choi et al., 1996) and fz (L. Zheng and R. Carthew, personal communication), which disrupt equator formation, display normal expression of mrr in the dorsal retina implying normal D-V patterning prior to the start of differentiation. Conversely, eyes in which the Hh pathway is ectopically activated can contain multiple equators, even when global patterning is not grossly altered (Chanut and Heberlein, 1995; Strutt and Mlodzik, 1995). Two lines of evidence suggest, however, that midline establishment and equator formation are linked mechanistically. First, we show here that Wg regulates gene expression along the D-V axis prior to and after the onset of differentiation, and when manipulations in wg expression alter the position of the D-V midline, the equator forms along the shifted midline. Second, previous analysis has revealed that mrr, expressed in the dorsal eye disc prior to the start of differentiation, participates in directing the position of the equator (McNeill et al., 1997). Our observation that wg controls mrr expression suggests that the effects of wg on D-V patterning and equator formation could be mediated, at least in part, by mrr.
Two possible models that relate D-V patterning and equator formation are diagrammed in Fig. 7C,D. The first model (Fig. 7C) postulates a causal temporal relationship between the establishment of the D-V midline and equator. The midline, defined through Wg signaling, would provide positional cues to generate an equatorial signal that directs WR122-lacZ expression and ommatidial rotation in an fz-dependent manner. As the fz-homolog Dfz2 has been implicated as a potential wg receptor (Bhanot et al., 1996), the equatorial signal may correspond to another Wnt family member. The equatorial signal, however, could be achieved indirectly through the definition of a punctual site where differentiation begins. In this scenario, discs containing more than one equator (Fig. 1K,L) could be generated as follows. Local overexpression of dpp (in ptc clones) along the disc’s margin may cause a local reduction of wg expression (Dpp is known to inhibit wg expression; Wiersdorff et al., 1996; Chanut and Heberlein, 1997; Pignoni and Zipursky, 1997), which in turn would generate an equatorial signal that propagates with the advancing MF. We suggested previously that the initial asymmetry created by the site of initiation might provide sufficient information to generate an equator (Chanut and Heberlein, 1995).
The second model (Fig. 7D) postulates that Wg signaling is involved in both the initial definition of the D-V midline and, independently and perhaps continuously, in spatially restricting the equatorial signal to the region of lowest Wg. In this model, the effect of Wg and the Fz-mediated equatorial signal would be opposite with regard to their effect on WR122-lacZ expression and ommatidial rotation. This model is supported by the observations that Wg is necessary to restrict WR122-lacZ to the equatorial region and that Wg is sufficient to alter ommatidial polarity (Treisman and Rubin, 1995; Tomlinson et al., 1997). The latter could be achieved indirectly, by interference with the equatorial signal, or directly, by guidance of ommatidial polarization. In the absence of more precise information about the time at which wg is required for its various D-V patterning effects, we cannot distinguish between these two models. Further, because these models are not mutually exclusive, it is possible that Wg acts early to help establish the equatorial signal (through definition of the D-V midline), and, later, to maintain this signal and direct ommatidial rotation.
ACKNOWLEDGEMENTS
We are very grateful to M. Mlodzik, A. Ephrussi, S. Cohen, and members of their laboratories at the EMBL in Heidelberg for hosting F. A. C. during part of this work. We thank M. Brodsky, S. Cohen, J. Treisman and the Bloomington Stock Center for various fly stocks; G. Rubin, K. Saigo and C. Goodman for antibodies; M. Wehrli, A. Tomlinson, D. Coen, L. Zheng and R. Carthew for sharing information prior to publication; D. Haumant for editing, and members of the Heberlein laboratory for discussions and support. The manuscript was greatly improved by comments from T. Wolff, R. Carthew, L. Zheng and J. Treisman. This work was supported by NIH grant EY11410 to U. H..