Combinatorial signaling by the Frizzled/PCP and Egfr-pathways during planar cell polarity establishment in the Drosophila eye

Summary

Frizzled (Fz)/PCP signaling regulates planar, vectorial orientation of cells or groups of cells within whole tissues. Although Fz/PCP signaling has been analyzed in several contexts, little is known about nuclear events acting downstream of Fz/PCP signaling in the R3/R4 cell fate decision in the Drosophila eye or in other contexts. Here we demonstrate a specific requirement for Egfr-signaling and the transcription factors Fos (AP-1), Yan and Pnt in PCP dependent R3/R4 specification. Loss and gain-of-function assays suggest that the transcription factors integrate input from Fz/PCP and Egfr-signaling and that the ETS factors Pnt and Yan cooperate with Fos (and Jun) in the PCP-specific R3/R4 determination. Our data indicate that Fos (either downstream of Fz/PCP signaling or parallel to it) and Yan are required in R3 to specify its fate (Fos) or inhibit R4 fate (Yan), and that Egfr-signaling is required in R4 via Pnt for its fate specification. Taken together with previous work establishing a Notch-dependent Su(H) function in R4, we conclude that Fos, Yan, Pnt, and Su(H) integrate Egfr, Fz, and Notch-signaling input in R3 or R4 to establish cell fate and ommatidial polarity.

Introduction

Planar cell polarization (PCP) is a critical patterning aspect in many organs and tissues and is manifest in all species analyzed. The orientation of cells (or groups of cells) within the plane of epithelia is established through the Wnt/Frizzled(Fz)-PCP pathway, generally referred to as Fz/PCP signaling. In Drosophila, most adult epithelial structures show PCP features including distal orientation of wing hairs, posterior orientation of bristles on the abdomen and thorax, and the regular arrangement of ommatidia in the eye (rev in Adler, 2002; Klein and Mlodzik, 2005; Lawrence et al., 2004; Mlodzik, 2002; Seifert and Mlodzik, 2007; Strutt, 2003). In vertebrates, examples of PCP include mammalian inner ear sensory epithelia, orientation of skin hair, or mesenchymal cell polarization in convergent extension during gastrulation (Guo et al., 2004; Keller, 2002; Montcouquiol et al., 2003). Several conserved core PCP factors regulate Fz-Dishevelled (Dsh)/PCP-signaling in all processes analyzed (rev. in Adler, 2002; Klein and Mlodzik, 2005; Mlodzik, 2002; Seifert and Mlodzik, 2007; Strutt, 2003).

Depending on the tissue context, Fz/PCP-signaling activates distinct effectors downstream of the core PCP genes. In the fly wing or mammalian inner ear, the primary PCP response is the reorientation of cytoskeletal elements, leading to localized actin polymerization (Adler, 2002; Eaton, 1997). In multicellular units like the ommatidia in the Drosophila eye, however, the primary response is regulation of cell fate specification via nuclear factors. PCP in the eye is established in third instar eye imaginal discs posterior to the morphogenetic furrow, where Fz/PCP-signaling leads to the differential specification of the R3 and R4 photoreceptors (rev. in Mlodzik, 1999; Strutt and Strutt, 1999; Wolff and Ready, 1993). Here Fz/PCP signaling downstream of Dsh includes Rho family GTPases, the STE20-like kinase Misshapen (Msn), and JNK/p38 type kinases (Boutros et al., 1998; Paricio et al., 1999; Weber et al., 2000). The AP-1 component Jun has been implicated as a transcription factor mediating the Fz/PCP cell fate induction (Weber et al., 2000). However, the mild phenotype of jun null alleles suggested other transcription factors act in R3/R4 cell fate specification downstream of Fz/PCP or other signaling pathways.

Jun is part of the transcription factor AP-1, consisting of either homo- or heterodimers of Jun and Fos family members (Eferl and Wagner, 2003; Jochum et al., 2001; Wagner and Eferl, 2005). In Drosophila, there is one Jun and one Fos gene (Kockel et al., 2001). Drosophila fos, called kayak (kay), is required for embryonic dorsal closure and fusion of imaginal disc epithelia during metamorphosis (Ciapponi et al., 2001; Riesgo-Escovar and Hafen, 1997; Zeitlinger et al., 1997). Analysis of fos has been difficult, as it functions downstream of several signaling pathways, including Egfr and JNK signaling (Ciapponi et al., 2001), and in imaginal discs loss of fos activity alters tissue growth or survival, and fos is required for normal progression through mitosis (Ciapponi et al., 2001; Riesgo-Escovar and Hafen, 1997; Zeitlinger et al., 1997; Hyun et al., 2006). Additional evidence for a role of Fos in cell morphology comes from C. elegans, where Fos is required for basement membrane removal during cell invasion in vulval development (Sherwood et al., 2005).

Members of the ETS transcription factor family have been shown to physically bind AP-1 or DNA target sites directly adjacent to AP-1 sites (Bassuk and Leiden, 1995; Chinenov and Kerppola, 2001). In Drosophila the ETS factors Yan/Aop (yan) and Pointed (pnt) act downstream of all receptor tyrosine kinase (RTK) signaling. The two factors act in opposition, Yan functioning as a repressor and Pnt as a transcriptional activator (Simon, 2000). Yan acts as a general repressor of photoreceptor fate induction (Rebay and Rubin, 1995), and has also been linked to Notch signaling (Rohrbaugh et al., 2002; Vivekanand et al., 2004). The potential role of these ETS factors has not been analyzed in the context of R3/R4 specification or PCP generation in the Drosophila eye.

Here we characterize the role of the transcription factors Fos, Yan and Pnt in R3/R4 specification. Using new Drosophila kay/fos alleles, we establish that Fos is required downstream (or in parallel) of Fz/PCP signaling in R3 to specify R3 cell fate. We demonstrate a specific requirement for the ETS factors Yan and Pnt in the PCP-dependent R3/R4 determination. In particular, Yan represses the R4 fate in R3 precursors, and Pnt appears to function downstream of Egfr-signaling to specify R4. As previous work has shown that Notch-dependent Su(H) is required to specify R4 (Fanto and Mlodzik, 1999; Tomlinson and Struhl, 1999), our data suggest that inputs from Fz/PCP, Notch, and Egfr-signaling are integrated at the nuclear level of the R3/R4 precursors via Fos (AP-1) and Yan function in R3, and Pnt and Su(H) in R4.

Materials and Methods

Flystrains and genetics

The kay/fos allele Df(3R)ED6315, kay^ED6315, has been generated following the Drosdel protocols (http://www.drosdel.org.uk/index.html), CB-6214-3 and 5-HA-1081 were used as starting RS lines and PCR to verify the extent of the deletion. Wing clones were generated with w, f^36a, hsflp; FRT82, arm-lacZ, f⁺, M/TM6b. Jun and fos double mutant adult eye clones were generated by crossing w, eyflp; FRT42, arm-lacZ/CyO; FRT82 kay²/TM3 with w; FRT42,jun^RC46/Sp;FRT82, arm-lacZ/TM6,Ubx and examining the white eye tissue, jun^IA109 was used in parallel. Adult eye clones of kay^P54 and kay¹⁶⁴⁴ were examined in w, eyflp; FRT82B kay⁻/FRT82B P[w⁺]D3 flies. Fos and fz disc clones were observed in eyflp; psq>GFP/+; FRT82B kay⁻/FRT82B arm-lacZ and eyflp; psq>GFP/+; fzR52 FRT80/arm-lacZ FRT80 larvae.

Flip-on dominant negative Egfr clones were generated by 8 minute heatshocks at 37^oC, 4–12 hrs prior to dissection in hsflp/+;act>y+>GAL4, UAS-GFP/UAS-DER^dn; GMR-p35/mδ-lacZ larvae. Flip-on Fos overexpression clones were generated by 1–2 1 hour heatshock at 37°C, 4 and 28 hrs prior to dissection in hsflp/+; act>y+>GAL4, UAS-GFP/; fos^EP[EY1271]/m∂-lacZ larvae. Hypomorphic kay² tissue was examined in an eyflp; FRT82 kay²/FRT82 arm-lacZ, M background. Yan function was analyzed in eyFLP; yan⁻ FRT40/ ubiGFP FRT40A; m∂-lacZ/+ discs.

Other fly strains used: pnt¹⁹⁸, pnt¹³¹⁸ (Ghabrial and Krasnow, 2006), pnt¹⁹⁰⁹⁹ (Weber et al., 2000), yan^XE18, yan^P (Lai and Rubin, 1992) and yan^E2d (Rogge et al., 1995), yan^E433 (Schober et al., 2005), pnt¹²³⁰ (Brunner et al., 1994), jun^RC46 (Kockel et al., 1997), kay¹, kay², tub-fos (Zeitlinger et al., 1997), UAS-fos (Ciapponi et al., 2001), fos^EP[EY12710], UAS-jun (Eresh et al., 1997), UAS-yan^EP[598] (Szeged stock center, http://expbio.bio.uszeged.hu/fly/), kay^P54 (Giesen et al., 2003), sepGal4 (Fanto et al., 2000), m∂-Gal4 (Gaengel and Mlodzik, 2003), m∂-lacZ (Cooper and Bray, 1999), sev-N^Δecd (Fortini et al., 1993). Other stocks are available from the Bloomington stock center (http://flybase.bio.indiana.edu/). Genetic interaction crosses with sev-Gal4[K25],UAS-pntP2 and yan^E2d/1; m∂-Gal4,UAS-GFP/+ animals were grown at 18°C. Other crosses were raised at 25°C.

Establishment of a new R3/R4 disc marker

To allow mosaic analysis of R3/R4 pairs in eye discs we established a new marker (psq^F112Gal4, UAS-GFP, referred to as psq>GFP), with high GFP expression in R3 and low GFP levels in R4 from ommatidial row 9 onward (Fig. 1B). psq>GFP was generated by converting psq^F112lacZ to a Gal4 insertion (Sepp and Auld, 1999) and recombining it with UAS-GFP. It was validated in fz mosaic eye discs. In fz mosaic R3/R4 pairs the wild-type cell adopts the R3 fate and ommatidial clusters chose its chirality accordingly. psq>GFP reflected these effects (Fig. 2B, E): Clusters, where R3 precursors had wild-type fz function, showed high psq>GFP expression in R3. Conversely if R4 precursors were wild-type for fz, most such clusters showed inverted chirality with psq>GFP high in R4 (see also Fig. 2B,E). psq>GFP analysis in fz mosaic R3/R4 pairs also revealed that the direction of rotation follows the cell fate identity of the fz⁺ cell (Suppl. Fig. S3). Similar to the fz analysis, expression of activated Notch in R3/R4 cells (sev-NΔecd, Fortini et al., 1993) changes many clusters to the R4-R4 symmetric type (not shown). psq>GFP reflected this reliably with the majority of clusters showing low R4-like expression in both cells of the R3-R4 pair. These studies establish psq>GFP as a faithful marker that represents the cell fate decision of R3 and R4 cells.

Figure 1. kay/fos is required for PCP establishment in the eye.

All panels are oriented with anterior left and dorsal up. (A) Schematic drawing of eye imaginal disc, showing establishment of PCP in developing ommatidial preclusters. Photoreceptors are depicted as ellipses with R3 and R4 precursors in green and light green, respectively. Cell fate specification of these two photoreceptors is a prerequisite for the establishment of chirality and direction of rotation. Ommatidial clusters rotate clockwise in dorsal half and counter clockwise in ventral half of disc, giving rise to two chiral forms in the adult eye (see schematic in right part of panel A), where the R3 rhabdomere takes a more polar position (green) and the R4 rhabdomere (light green) sits closer to R7. (B,B’) Chirality and rotation in a wild-type disc assayed by a novel marker for photoreceptors R3 and R4: psq-Gal4, UAS-GFP (green, and monochrome in B’) in R3 at high levels and R4 at low levels, allowing identification of both cell fates from ommatidial row 9 on. Anti-Elav (blue; all R-cells) and anti-Bar (red; R1/R6) are also shown. (C) Wild-type adult eye showing the two ommatidial chiral forms, lower panel depicting them schematically as black and red arrows (separated by the equator in yellow).

(D) Mutant eye clones of kay^P54 show rotation and chirality defects. Loss of redish pigment marks mutant tissue, with schematic representation on the right (arrows are as in panel C; ommatidia that have lost photoreceptors are indicated by black dots). Eyes displayed an average of 13% chirality defects, 10% rotation defects and 17.3% loss of photoreceptor/survival phenotypes. % based on counting ommatidia of 8 eyes in mutant tissue.

Figure 2. kay/fos is required for R3/R4 cell fate establishment in developing eye discs.

(A-D) Confocal pictures of 3rd instar eye discs. Photoreceptors are marked with anti-Elav (blue), and the R3 and R4 cells by psq>GFP (green; monochrome in right panels). Note high levels of psq>GFP in R3 and low levels in R4 (some examples are marked with “3” in green and “4” in yellow). (A) wild-type disc. (B-D) Clones of indicated genotype, mutant tissue is marked by loss of red marker (arm-lacZ). (B) fz^R52 mosaic eye disc. Note that in fz⁻ mosaic R3/R4 pairs the wild-type cell always expresses GFP at high levels, presumably adopting R3 fate, and the ommatidial cluster rotates according to that decision (see Suppl. Fig. S3 for quantification of rotation behavior). Some examples that show either switched fates or symmetrical clusters are indicated by green [R3 fate] or yellow [R4 fate] asterisks. Within fz mutant clones ommatidial clusters can also express the R3/R4 specific psq>GFP marker in more than 2 cells per cluster (examples indicated by orange arrowheads). (C-D) Expression of the psq>GFP marker in kay mutant ommatidia reflects R3/R4 fate and chirality defects; (C) kay¹⁶⁴⁴ and (D) kay^ED6315. Specific features are highlighted as in (B). (E) and (F) Quantification comparing fz^R52 null and kay^ED6315 hypomorphic mutant effects in mosaic R3/R4 pairs. The graphs summarize the analysis of mosaic R3/R4 pairs in eye discs and the effect on chirality establishment. If R3 precursors were wild-type and R4 precursors mutant for fz (E) or kay/fos (F) 84% of R3/R4 pairs expressed psq>GFP correctly. If R3 was fz mutant and R4 wild-type, 78% of the pairs showed inverted psq>GFP expression (E), which is in accordance with previously published analysis in adult eyes (Zheng et al., 1995). (F) In kay^ED6315 mosaics with a mutant R3, 16% of clusters show inverted expression (6.7% for kay^P54) and 50% show equal, low expression (27% for kay^P54, not shown), indicating that kay/fos is required in R3. Note that such clusters are underrepresented (n=38) as compared to the opposite mosaic combination (n=63), which confirms a genetic requirement of kay/fos in R3 not only in PCP specification.

Immunohistochemistry and histology

Cuticle preparations, antibody stainings, and adult eye sections were performed as described in Drosophila Protocols [William Sullivan (Editor), M. Ashburner, R. Scott Hawley]. Antibodies used were: Rat-anti-Elav (1:50; from DHSB), rabbit-anti-Fos (1:500; from D. Bohmann), mouse-anti-βgal (1:100; Promega), rabbit-anti-βgal (1:1000; Cappel), guineapig-anti-Sensless (1:1000: from H. Bellen) and rabbit-anti-Bar (1:100, K. Saigo). Confocal laser scanning microscopy was performed at the MSSM Microscopy Shared Resource Facility.

Results

Fos is required for planar cell polarity in the Drosophila eye

PCP establishment in the eye is mediated by an interplay of Fz/PCP and Notch signaling in the R3/R4 precursor pair, leading to their correct specification (Fig. 1A-C; Cooper and Bray, 1999; Fanto and Mlodzik, 1999; Tomlinson and Struhl, 1999). The precursor that is closer to the midline (equator) of the eye field - the future R3 - receives higher Fz-activity and will be specified as R3. This leads to increased transcription of the Notch ligand Delta (Dl) in R3. Dl then activates Notch in the neighboring precursor, inducing it as R4. Subsequently, the clusters rotate 90° in each half (clockwise or counterclockwise) towards the midline. The R3/R4 cell fates are later translated into distinct chiral ommatidial forms in the dorsal and ventral halves of adult eyes through associated cell rearrangements (Fig. 1A-C; Fanto and Mlodzik, 1999; Mlodzik, 1999; Tomlinson and Struhl, 1999; Wolff and Ready, 1993). In mutants of fz or any component of Fz/PCP-signaling these processes are perturbed, reflected in incorrect R3/R4 fate specification, chirality defects, and ommatidial misorientation. Although Fz/PCP signaling regulation is well studied, little is known about the nuclear events acting downstream of Fz/PCP signaling in R3/R4 fate determination (Klein and Mlodzik, 2005; Strutt, 2003).

Genetic interactions with a PCP gain-of function (GOF) genotype, sev-Fz, have suggested that kay/fos might act in R3/R4 specification (Ciapponi et al., 2001). This is supported by the observation that Jun, the heterodimeric partner of Fos generates mild PCP eye phenotypes when compromized (Weber et al., 2000). Since clones of the hypomorphic kay² allele show no phenotype and jun null alleles reveal only a mild PCP phenotype in the eye (see below and Weber et al., 2000), we tested for redundancy and attempted to generate double mutant clones of kay² and a jun. Strikingly, double mutant cells were not recovered (see Methods), suggesting that jun and fos share a partially redundant requirement in imaginal discs (not shown).

Null mutant kay/fos clones anterior to the morphogenetic furrow (MF) in actively dividing cells were not recovered (Suppl. Fig. S1A) supporting previous observations that kay/fos has a requirement in cell cycle progression and proliferation (Hyun et al., 2006). Therefore, to analyze developmental patterning and PCP requirements of kay/fos, we identified new alleles that did not affect proliferation and survival but were stronger than kay². We characterized two P-element induced alleles, kay^P54 (Giesen et al., 2003) and kay¹⁶⁴⁴ (Flybase) and the deletion allele Df(3R)ED6315 (subsequently referred to as kay^ED6315) that removes 3 of the 4 alternative 5’ coding-exons (Suppl. Fig. S1E; Methods). All three alleles led to reduced Fos protein levels in homozygous mutant tissue (Suppl. Fig. S1C, D and not shown) and caused typical kay/fos LOF phenotypes in embryos and imaginal disc fusion (Suppl. Fig. S2, and Tables S1 and S2). Expression of exogenous fos (tub-fos) or loss of one genomic copy of the JNK-specific antagonist puckered (puc) rescued embryonic lethality of these alleles in trans to kay¹ or kay² (Suppl. Table S1 and Fig. S1). Our phenotypic and genetic analyses indicate the following allelic series according to their strength: kay¹> kay^ED6315>kay^P54>kay¹⁶⁴⁴>kay². Importantly, these new alleles permitted cell proliferation/survival in imaginal disc tissue and allowed us to analyze the role of kay/fos in PCP establishment.

kay^P54 and kay¹⁶⁴⁴ mutant clones showed ommatidia with inverted or lost chirality (symmetrical clusters), rotation defects and loss of photoreceptors in adult eyes (e.g. Fig. 1D). Reducing fos function in the developing wing showed no PCP defects (>10 pupal and adult wings with kay^ED6315, kay^P54 or kay¹⁶⁴⁴ clones were analyzed, not shown). Similarly, no wing PCP phenotypes were found associated with jun⁻ clones (not shown). Thus kay/fos appears to be a tissue specific candidate to the nuclear PCP response in R3/R4 specification in the eye (see below).

Fos is required for photoreceptor R3/R4 fate specification

To ask which cell within the R3/R4 pair required kay/fos function, we analyzed kay mosaic tissue. We focused on developing eye discs at the stage of PCP establishment, because at this stage all photoreceptor cells are still present in kay mutant tissue. We generated a new marker (psqF112Gal4, UAS-GFP, further referred to as psq>GFP, see Methods) for R3/R4 precursor fate analysis. R3 expressed high GFP while R4 expressed low levels (Fig. 1B and 2A). The psq>GFP marker was validated in fz⁻ mosaic eye discs, and it precisely reflected the known features of fz mosaic ommatidia (Fig. 2B, 2E; see Methods for details and supp. Fig S3) in accordance with adult mosaic fz analysis (Zheng et al., 1995).

Mutant tissue of the new kay alleles showed altered expression of psq>GFP in ommatidial preclusters (Fig. 2C,D), similar to fz (Fig. 2B), with kay^ED6315 showing the strongest effect (Fig. 2D). In such kay mutant clones the psq>GFP level in R3/R4 precursors is often inverted (high in the R4 position as compared to R3) or equal (symmetrical) in both cells. Thus alterations in psq>GFP expression are reflecting chirality inversions and symmetrical clusters as observed in adult eyes (Fig. 1D and not shown).

In kay/fos mosaic R3/R4 pairs, where the R3 cell is wild-type and R4 mutant, the majority (85.7%) displayed correct marker expression (Fig. 2F), very similar to fz mosaic clusters with wild-type R3 (Fig. 2E). In contrast, when R3 precursors had lost kay/fos function, aberrant psq>GFP expression was observed for kay¹⁶⁴⁴, kay^P54, and kay^ED6315 alleles (Fig. 2C,D, not shown). In most cases we observed equal low, R4-like expression in both cells in mosaic pairs (Fig. 2C,D and F; see legend Fig. 2 for details). The weaker effect of kay/fos mutants as compared to fz (Fig. 2E-F) are likely due to the fact that the kay LOF alleles are hypomorphic and still retain some function; alternatively, kay/fos may act redundantly with jun in this context. Comparing the adult chirality defects in kay^P54 to the disc phenotype revealed somewhat lower percentages, presumably because the loss of photoreceptors obscures chirality defects in the adult (Fig. 1D, 2F).

In summary, the observation that kay/fos LOF in R3 invert or equalize psq>GFP expression within the R3/R4 pair indicates that kay/fos is required cell autonomously to establish the R3 fate.

Fos is sufficient to induce R3 within the R3/R4 pair

We next tested if exogenous Fos expression is sufficient to interfere with PCP establishment. Whereas tub-fos expression could rescue the kay^1644/1 eye phenotype (Fig. 3A), increased Fos levels (generated via tub-fos) in a wild-type background were sufficient to cause misrotations and chirality defects. In otherwise wild-type eye discs, tub-fos caused inverted psq>GFP expression in R3/R4 pairs (Fig. 3C). Similarly, overexpression in R3/R4 pairs with sevGal4 caused dramatic effects, leading to an average 40% of adult ommatidia with a symmetric R3/R4 arrangement (Fig. 3B). These data suggest that relative levels of Fos within the R3/R4 pair are critical for PCP establishment. To determine which cell of the R3/R4 cells is sensitive to elevated Fos levels, we overexpressed Fos in clonal patches. When Fos was overexpressed in R4 precursors, the R4-specific m∂-lacZ expression was lost in the majority of clusters or the R3 precursor would show expression (Fig. 3D). Fos overexpression had no effect in R3 (Fig. 3D), confirming that the R4 precursor is sensitive to elevated Fos levels. These data further indicate that kay/fos is instructive for the R3 fate, and that differential Fos levels within the R3/R4 pair are important for correct fate induction. Fos overexpression had no effect on PCP features in other tissues including the thorax or wing (not shown).

Figure 3. Fos is sufficient to induce PCP defects in the eye.

(A-B) Adult eye sections, with respective schematic representation in lower panels (arrows as in Fig. 1, symmetric ommatidia are indicated by green arrows). (C) Developing eye disc marked as in Fig. 2. (A) Wild-type phenotype as seen in tub-fos/+ rescue of kay¹⁶⁴⁴/kay¹. (B) sev-Gal4/+, fos^EP[EY12710]/+ eyes show on average 40% symmetrical and 3% inverted chirality (n=394) at 29°C. (C) In tub-fos/Y eye discs stained for Elav (blue) and psq>GFP (green), chirality defects are apparent as indicated (green stars: high level GFP, R3 fate and yellow stars:low level GFP, R4), indicating inverted chirality. (D) Summary of Fos overexpression in single R-cell precursors. Panel lists which R-cells expresses Fos exogenously and the effect this has on R4 fate (m∂-lacZ marker). Only the R4 precursor of the R3/R4 pair is sensitive to elevated Fos levels loosing m∂-lacZ expression and R4 fate.

Together, the overexpression and LOF results indicate that Fos is required cell-autonomously for R3 fate specification and it is both necessary and sufficient to convey that fate and PCP establishment within the eye. Therefore it behaves like a core component of Fz/PCP signaling.

The ETS transcription factors Yan and Pnt interact with Fos/Jun and show PCP defects

The ETS transcription factors Yan/Aop (yan/anterior open) and Pointed (Pnt) act antagonistically downstream of receptor tyrosine kinases (RTKs) during photoreceptor induction and other developmental processes (Simon, 2000; Vivekanand et al., 2004). We isolated yan and pnt alleles as dominant modifiers of an ectopic Jun-dependent eye PCP phenotype (Weber et al., 2000), and unpublished data), suggesting a functional link between these factors and AP-1 (Fos/Jun) activity during PCP establishment. We therefore tested whether yan and pnt can modulate eye phenotypes of kay² and jun⁻.

Strikingly, removing one genomic copy of yan in homozygous kay² mutant tissue led to the emergence of chirality defects apart from the expected loss of photoreceptors (Fig. 4B). This enhancement was observed with different allelic combinations (see Fig. 4 legend). By itself, jun⁻clones show only mild phenotypic defects (on average 2% of ommatidia display PCP defects; (Weber et al., 2000). In a pnt heterozygous background, however, the PCP defects of jun⁻ clones are enhanced (Fig. 4C; both PCP and photoreceptor number defects are increased compared to control jun⁻ clones as observed for several allelic combinations, see Fig. 4 legend). We therefore conclude that yan and pnt have the potential to cooperate with the AP-1 factors in establishing the R3/R4 fates.

Figure 4.

yan and pointed affect polarity establishment in the eye and cooperate with AP-1 in this process. (A-C) Adult eye sections with respective schematic representations in lower panels (marked as in Fig. 3). (D, F) Confocal pictures of eye discs of genotypes indicated stained with anti-Elav (red: all photoreceptors), psq>GFP (green R3/R4 in D) and m∂-lacZ (green R4 in F; monochrome in lower panels). (B-F) For quantification of defects and further genotypes analyzed see Methods. (A) Wild-type phenotype of mostly kay² mutant tissue indicated by loss of pigment. (B) kay² clones in yan^E2d/+ mutant background display chirality, rotation and photoreceptor number defects. Note that kay² clones do not show a PCP phenotype by themselves (panel A). (C) jun^RC46 clones in a pnt¹⁹⁰⁹⁹ heterozygous background display increased number of PCP phenotypes as compared to jun clones alone. kay²/fos clones were generated in yan^−/+ and jun⁻ clones in pn^−/+ because of their chromosomal location. kay² clones in yan^E2d/+ showed 31.5% defective ommatidia and 4.1% with aberrant chirality. 539 ommatidia in 8 eyes were evaluated. jun clones in pnt heterozygous background displayed increased number of PCP phenotypes as compared to jun clones alone. 9.4% chirality defects were observed in 328 ommatidia in mosaic areas from 6 eyes, jun^RC46, jun^IA109 and pnt^d88 or pnt¹⁹⁰⁹⁹ alleles were used. kay^P54 clones in yan^E2d/+ showed too many photoreceptor number defects and were thus not scorable; kay² clones in yan^1/+ showed a mild effect.

(D,D’) pnt^1277/1230 mutant disc; such discs displayed 10% inverted and 9% symmetric clusters as indicated by psq>GFP expression levels (asterisks are as in Fig. 3A-D). The psq>GFP marker highlights clusters with inverted or even expression levels as indicated by asterisks. Clusters with more than 2 cells expressing the marker were not scored for chirality. The orange doted line indicates the equator. pnt^1277/Δ88 discs showed 6.7% inverted and 6% symmetrical clusters with psq>GFP (not shown). m∂-lacZ, serving as a R4 specific marker confirmed PCP defects in pnt^1230/Δ88 and pnt^KG04968/Δ88 discs, with 26.4% and 9.5% chirality defects, respectively (not shown).

(E) Graph summarizing chirality defects observed in yan^1/E2d mutant discs stained for the photoreceptor R4 fate with m∂-Gal4,UAS-GFP. Clusters were counted according to lost marker expression reflected by yellow bar, equal R3 and R4 expression by orange and R3 expression by red bar. Discs were co-stained for all photoreceptors (Elav) and R8 (Senseless). Control yan/+ discs had 6.9% defects. yan^1/P discs showed 9.4% chirality defects as monitored by psq>GFP, controls had 3% defects (not shown).

(F) Example of a mosaic yan^XE18 disc, wild-type tissue is marked in blue. Two clusters, which have lost yan function in R3 (indicated by arrowheads) express the R4 specific marker m∂-lacZ in R3, compare Table 1A. If several photoreceptors in a cluster loose yan function, multiple cells express m∂-lacZ (see upper right area). For graphs and percentages the sum of 3 discs and 110–428 ommatidia were evaluated for each genotype.

We next asked whether mutant tissue for pnt or yan alone shows PCP defects. The respective null mutants cannot be scored due to effects on photoreceptor induction, differentiation, and survival, and so instead we analyzed hypomorphic allelic combinations of pnt and yan in eye discs. The pnt^1277/1230 and pnt^1277/Δ88 loss-of-function combinations showed chirality defects as assessed by psq>GFP expression (Fig. 4D); similar defects were observed with the R4 specific m∂-lacZ reporter in pnt^1230/Δ88 and pnt^KG04968/Δ88 discs (see Fig. 4 legend; not shown). The yan allelic combinations yan^1/E2d and yan¹/^P also showed consistent PCP defects (Fig. 4E and not shown). Taken together, these analyses indicate that pnt and yan are both required for correct establishment of the R3/R4 fates and ommatidial chirality.

Different signaling pathway interactions of Fos/Jun and Pnt and Yan in PCP establishment

The above data established that Pnt and Yan are required for R3/R4 specification. Their modulation of the jun and kay² phenotypes suggests that they cooperate with the AP-1 factors in this context. Previous work has suggested that in R3/R4 specification Jun acts downstream of Fz, Dsh and JNK signaling (Boutros et al., 1998; Paricio et al., 1999; Weber et al., 2000). We thus examined the genetic behavior of Fos, Pnt and Yan relative to a R3/R4 specific Dsh gain-of-function phenotype.

Dsh overexpressed in R3/R4 precursors (under sev-enhancer control; sev-dsh; Boutros et al., 1998) caused 42% of adult ommatidia to acquire a symmetric arrangement (Fig 5A,D). Reducing the genomic copy number of downstream effectors in such sensitized backgrounds can suppress the respective phenotypes (reviewed by St Johnston, 2002). Of the four transcription factors tested, the strong kay/fos alleles and jun alleles suppressed sev-dsh (Fig. 5B,D), and restored the equator (Fig. 5B). Alleles of yan mildly suppressed but did not eliminate inverted chirality (Fig. 5C,D). pnt alleles did not modify sev-Dsh, even when the associated cell-death, reflected by loss of photoreceptors, was inhibited by GMR-p35 (Fig. 5D, see legend for details).

Figure 5.

kay/fos alleles dominantly suppress sev-Dsh. (A-C) Eye sections (upper panels) and schematic representations (lower panels; arrows as in Fig. 4). (A) 2xsev-dsh/+ caused on average 42% symmetric ommatidia and inverted chirality. (B) 2xsev-dsh/+, kay¹/+ eyes: occurrence of symmetric ommatidia is reduced to 16%, no inverted chirality is observed and the equator is restored (note that the latter effects of suppression are not represented in graph, panel D). (C) yan^E2d/+; 2xsev-dsh/+ eyes show 28% symmetric ommatidia. The equator is absent. (D) Quantification of interactions with sev-dsh as scored by counting symmetric ommatidia only. Reducing jun function (control, Boutros et al. 1998) or kay/fos by one copy suppresses all chirality defects in sev-dsh and restores the equator. yan reduces symmetrical ommatidia formation, but clusters with inverted chirality are still present and no equator can be determined. pnt caused loss of photoreceptors in sev-Dsh, which precluded the analysis. When cell death was inhibited by GMR-p35 it was apparent that pnt had no effect (~400 ommatidia from 3 eyes were counted each).

Taken together with the cellular requirements of jun and fos (see above and Weber et al., 2000), these data suggest that jun and fos might function as downstream effectors of Fz/Dsh-PCP signaling (or in parallel to it; see Discussion) to specify R3 and to set up chirality in the eye. In contrast, the role of pnt is likely downstream of other signaling input (see below). Yan can be regulated by several MAPK members (JNK and ERK type) and thus its mild interaction with sev-Dsh suggests that it could act in Fz/Dsh and other pathways (see below).

Egfr is required to establish the R4 fate

Pnt and Yan are known to act downstream of Egfr-signaling in several contexts in the developing embryo and eye (rev. in Casci and Freeman, 1999). It is however not possible to analyze strong LOF Egfr alleles in the R3/R4 context due to the reiterative Egfr requirements in eye development (Casci and Freeman, 1999). Therefore, we first tested for a potential Egfr-requirement in PCP by analyzing genetic interactions between fz and Egfr mutant alleles (Fig. 6A,B). Reduction of a genomic copy of Egfr in a fz^P21/fz^J22 hypomorphic background suppressed the fz LOF PCP eye phenotype (Fig. 6B). In addition, the hypomorphic Egfr^3C81/Egfr^top1 combination showed mild PCP defects, which were suppressed by a reduction in a genomic copy of fz (Fig. 6A). Furthermore, such antagonistic interations between fz and Egfr were also observed in double heterozygous scenarios: a single copy of an Egfr null allele (Egfr^CO/+) shows a mild dominant (haplo-insufficient) PCP phenotype reflected by misoriented ommatidia, which is suppressed in the presence of fz heterozygosity (fz^P21/+ or fz^R52/+; see Fig. 6 legend). These data indicate that Egfr-signaling functions in the R3/R4 fate decision and PCP establishment and that there is a required balance between Fz/PCP and Egfr-signaling.

Figure 6. Egfr is required for R4 fate specification.

(A-B) The fz and Egfr associated PCP phenotypes are suppressed by dosage reduction in the other gene. (A) Reducing a genomic copy of fz (fz^P21 and fz^R52) suppresses the defects of Egfr mutant eyes as indicated by increased percentage of correctly oriented wild-type ommatidia. (B) Reducing a genomic copy of Egfr suppresses the occurrence of symmetric ommatidia in fz mutant eyes. 3–4 eyes were counted and ~400 ommatidia evaluated for each genotype. Egfr fz mutant analysis was also evaluated for top^CO/1 and top^CO/+ with fz^P21/+ and fz^R52/+. In both cases fz dominantly suppressed the rotation defects and loss of R-cells. In top^CO/+, rotation defects were suppressed by fz heterozygosity from 26% to 12% or 18%, respectively.

(C-F) Confocal pictures of 3rd instar eye discs. Photoreceptors are marked with anti-Elav in red. (C) sep-GAL4, UAS-Egfr[DN]/+; m∂-lacZ/+ disc. Examples of 5 cell ommatidial preclusters which lost m∂-lacZ expression are marked with arrowheads (16.8% of total ommatidia. Note that only clusters with the full complement of R-cells were counted). R8 is marked with anti-Sens (blue) and R4 fate with m∂-lacZ (green). (D-F) Mosaic tissue containing cells expressing dominant negative Egfr marked by GFP (green) and the R4 marker m∂-lacZ (blue). White arrowheads mark R4 precursors, yellow arrowheads R3 precursors. Note that R4 cells expressing Egfr[DN] loose m∂-lacZ in R4 and then R3 expresses m∂-lacZ (arrowheads in D and E). Egfr[DN] expression in R3 (or other R-cells) does not affect R4 fate (arrowheads in F, quantif. in G). Egfr[DN] expression in both R3 and R4 cells leads to loss of m∂-lacZ (orange arrow in E). (G) Summary of mosaic analyses with Egfr[DN] in ommatidial preclusters. Individual R-cells positive for Egfr[DN] in otherwise normal 5 cell preclusters were counted and the m∂-lacZ R4 fate marker was registered. Only the R4 precursor was sensitive to Egfr[DN].

To bypass the general Egfr-requirements in eye development, we used expression of a dominant negative (DN) isoform in a temporal and space-restricted manner; an approach used successfully to dissect several of the specific roles of Egfr in eye development (Freeman et al., 1996; Brown and Freeman, 2003). Low level Egfr^DN expression in photoreceptors R3 and R4 during PCP establishment (by means of the sev enhancer-promoter driver; Fanto et al., 2000) caused loss of m∂-lacZ expression and by inference R4 fate (Fig. 6C). Next we asked which cell required Egfr function in PCP establishment with single cell expression clones of Egfr^DN (see Methods) and monitored cell fate with m∂-lacZ R4 marker in otherwise wild-type preclusters. Since Egfr-signaling is required for photoreceptor survival (Freeman and Bienz, 2001), cell death was prevented by co-expressing Egfr^DN with p35. Strikingly, in 50% of clusters where the R4 precursor had reduced Egfr-activity (expressing Egfr^DN), the m∂-lacZ R4 marker was abolished or was turned on ectopically in R3, indicating that the cluster has acquired inverted chirality (Fig. 6D,E and G). Equivalent Egfr^DN clones with expression in R3 (or other photoreceptor precursors) did not display an effect on the R3/R4 fate decision (Fig. 6F,G).

Taken together with the sev-driven Egfr^DN expression, our mosaic analysis demonstrates that high Egfr-signaling levels are required in R4 to specify its fate relative to R3. These data, together with previous work addressing Notch signaling, establish a combinatorial signaling input into R4 fate specification, requiring both Notch activation (mediated through Fz induction of Dl in R3) and Egfr-signaling. They also suggest that there is a requirement for a tightly regulated balance between Fz/PCP and Egfr-signaling in the R3/R4 pair (see also below).

Yan acts in R3 to repress the R4 fate

Next we analyzed the cellular requirements of the Egfr-signaling effectors Yan and Pointed (Pnt). Utilizing yan and pnt alleles that permitted the 5 cell ommatidial preclusters to form normally, we asked which cell of the R3/R4 pair required pnt and yan function for proper PCP establishment. In the majority of yan mosaics (using yan^XE18; Lai et al., 1997, and yan^E433; Schober et al., 2005) where the R3 cell had lost yan function, ectopic R4 specific mδ-lacZ expression was detectable in R3, indicating that this cell had acquired the R4 fate. Conversely, single mutant R4 cells showed no defects in R3/R4 fate induction (Fig. 4F; summ. in Table 1A). These data indicate that Yan is required in R3 to repress the R4 fate.

Table 1.

Mosaic analysis of yan and pointed mutants.

A
Allele	R3 mutant	R4 mutant	R2 mutant	R5 mutant	R8 mutant
yanE433
wild-type (m∂ in R4)		11	3		1
mutant (m∂ in R3+R4)	4			1
n	4	11	3	1	1
yanXE18
wild-type (m∂ in R4)	11	17	5	3
mutant (m∂ in R3+R4)	12
n	23	17	5	3	0
total n	27	28	8	4	1
effect/conclusion	yan represses R4 fate in R3	yan function is dispensable in R4	under represented genotype	under represented genotype	under represented genotype
B
Allele Marker	R3 mutant	R4 mutant	R2 mutant	R5 mutant	R8 mutant
pnt1230 psqGAL4, UAS- GFP
wild-type (R3 high, R4 low)	3	5	1	1
mutant (symmetrical) (inverted)	3 1	1	1
n	7	6	2	1	0
pnt19099 m∂-lacZ
wild-type (m∂ in R4)	1				3
mutant (symmetrical) (inverted)	1	1	2 1	2	1
n	2	1	3	2	4
pnt198 m∂-lacZ
wild-type (m∂ in R4)					4
mutant (symmetrical) (inverted)	2 1	2		3	1
n	3	2	0	3	5
total n	12	9	5	6	9
wild-type	4	5	1	1	7
mutant (symmetrical) (inverted)	5 3	4	3 1	5	2
effect	loss of R4 fate or inverted R3/R4 fates	loss of R4 fate	loss of R4 fate or inverted R3/R4 fates	loss of R4 fate	loss of R4 fate

Summary of single mutant cell analysis in 5 cell preclusters in developing mosaic eyes. R3, R4, R2, R5 and R8 were scored for loss of gene function and R4 fate was monitored with m∂-lacZ. (A) For yan^E433 and yan^XE18 mosaic clusters were counted in yw eyFLP; yan- FRT40/ ubiGFP FRT40A; m∂-lacZ/+ eyes. If yan was mutant in the R3 cell 59.3% of these cells adopted the R4 fate (n=27). No effect was observed in R4 (n=28). Clones for R2, R5 and R8 were underrepresented due to the general function of yan as a repressor of differentiation and the sequential recruitment of R-cells to the cluster.

(B) For pnt¹²³⁰, pnt¹⁹⁰⁹⁹ and pnt¹⁹⁸ mosaic clusters were counted as described above. For all 3 alleles single cell clones for R2, R5, R3 and R4 abolished the R4 cell fate and/or changed the R3 fate to R4. Clones for R8 showed mostly no effect.

To assess pnt we examined mosaic discs containing single cells mutant for pnt¹²³⁰, pnt¹⁹⁰⁹⁹ and pnt¹⁹⁸ (Ghabrial and Krasnow, 2006; Table 1B). Partial pnt loss in any of the R3, R4, R2 and R5 cells often abolished the R4 fate (loss in R8 rarely had an effect). We conclude that mosaic analysis of pnt mutants in the developing eye shows a requirement of pnt for the R4 fate, and that this is in part a non cell-autonomous effect due to the feed-back loops of Egfr-ligand expression (see Discussion).

Taken together, our data suggest that the ETS transcription factors pnt and yan are required for R3/R4 fate specification, with yan displaying a R3-specific requirement and pnt required for R4. As they dominantly enhanced kay and jun phenotypes, they cooperate with the AP-1 factors jun and fos in this process to integrate input from Fz/PCP and Egfr-signaling pathways.

Pnt specifies R4 and is antagonized by Yan

yan exerts its function within R3 (Fig. 4F, Table 1A), but the mosaic analysis of pnt was complicated by its requirement in the ligand feedback loops. Therefore we asked whether overexpression of Pnt (the PntP2 isoform, a general effector of Egfr-MAPK/Erk signaling; Brunner et al., 1994; O’Neill et al., 1994; Rebay and Rubin, 1995; Rogge et al., 1995) would provide additional insight. Strikingly, sevGal4 driven PntP2 yielded a majority of ommatidia with a symmetrical arrangement (Fig. 7A,C). Examining the developing eyes with R3 and R4 specific markers, Dl-lacZ and m∂-lacZ, revealed that the R3/R4 cell pair had lost their differential identity at the expense of many R4/R4 symmetrical clusters (revealed by the expression of the R4-specific mδ-lacZ marker in both cells; Fig. 7B). Thus PntP2 is sufficient to induce R4 fate (Fig. 7B). These data support the importance of a proper balance of Pnt activity in the R3/R4 cells.

Figure 7. PntP2 is sufficient to induce R4 fate. yan and jun antagonize pntP2.

(A) sev-GAL4, UAS-pntP2/+ eye section with schematic representation on the right (arrows drawn as in Fig. 1). Symmetric ommatidia are represented by green arrows. sev-GAL4, UAS-pntP2/+ eyes showed 70% symmetric ommatidia. (B) Many ommatidial preclusters in sev-GAL4, UAS-pntP2/+ show both cells of the R3/R4 pair expressing m∂-lacZ, the R4 fate marker. Elav indicates all photoreceptors in red. (C) Graph summarizing average defects of over expression combinations. Ommatidial clusters were scored for wild-type (black), symmetric (green) and wrong (inverted) chirality (grey). Genotypes are indicated below graph. Yan antagonizes the PntP2 effect, and Jun does mildly. 400–500 ommatidia of 4 eyes were evaluated each. Single over expression of wt Jun, Fos and Yan did not cause PCP defects at 18°C and 25°C. (C) Model of combinatorial input into R3/R4 specification (see text for details). Note that lower levels of Egfr and Pnt are required for cell survival/induction of all R-cells (indicated by smaller grey Egfr writing).

Co-expression of Pnt and Yan led to a suppression of the gain-of-function PntP2 phenotype, as anticipated (Fig. 7C). Interestingly, Jun co-expression was also able to antagonize the PntP2 effect, reducing the frequency of symmetric clusters and increasing % of wild-type chirality (Fig. 7C). Fos did not show an effect (see legend to Fig. 7 for controls).

These experiments indicate that Yan can antagonize Pnt in the R3/R4 decision, consistent with their functional relationship in other contexts (Rohrbaugh et al., 2002; Vivekanand et al., 2004). The effect of co-expressing Jun and Pnt suggests that they also act antagonistically in this context. These data provide further functional support for a pnt requirement in R4, as both yan (this study) and jun mosaics indicate that they are required in R3 for R3 fate induction (Weber et al., 2000). Taken together with the autonomous Egfr-signaling requirement for R4 fate, we conclude that pnt acts downstream of Egfr-signaling in R4 to specify the R4 fate.

Discussion

Here we demonstrate a specific requirement for Egfr-signaling and the transcription factors Fos (AP-1), Yan, and Pnt in PCP signaling dependent R3/R4 specification. Loss and gain-of-function assays indicate that these transcription factors integrate input from Fz/PCP and Egfr-signaling, which together with Su(H) function downstream of Notch-signaling leads to correct R3 and R4 specification. Our data show that (1) Fos is required in R3, specifying its fate (similar to Fz), (2) Egfr-signaling functions in R4 for its specification, (3) Yan is required in R3 to inhibit R4 fate, and (3) Pnt functions for R4 specification (probably downstream of Egfr-signaling). The signaling network is completed by Notch-signaling and its nuclear effector Su(H) functioning to specify R4. Thus a combinatorial input from multiple signaling pathways is reflected by the requirements of distinct transcription factors that lead to correct R3/R4 induction and ommatidial polarity (see Fig. 7D).

R3 specification

Previous studies established that Fz is required cell-autonomously for R3 fate induction (Zheng et al., 1995). Our analyses of kay/fos LOF alleles indicate that Fos is also required cell-autonomously in R3 for its fate determination. When overexpressed, Fos also acts like Fz in R3/R4 photoreceptors at the time of PCP establishment, with the cell of the pair that has higher Fos levels adopting the R3 fate. Based on its requirement in R3 and genetic interactions, Fos could act as a nuclear effector of Fz/PCP signaling. This is supported by the observation that it is able to suppress sev-dsh induced PCP defects (Fig. 5; the genetic data can however not rule out that Fos could act in parallel to Fz/Dsh-PCP signaling). The subtle differences observed between fz and kay/fos LOF requirements - in fz⁻ R3/R4 mosaics the wild-type cell adopts the R3 fate often causing chirality inversions, while in kay/fos mosaic pairs with a mutant R3 the pair often adopts symmetrical R4/R4 appearance – is likely due either to the hypomorphic nature of the kay/fos alleles that had to be used in the analysis or potential redundancy with jun (see below).

In addition to the positive Fos signaling input, R3 specification also requires the repressor function of Yan, with Yan inhibiting R4 fate in the R3 precursor (Fig. 7D). This is evident by the cellular requirement of Yan and highlighted by the increased defects in a kay/fos and yan double mutant scenario, where both aspects are partially impaired causing frequent R3/R4 fate decision defects. The dominant enhancement of kay² by yan LOF suggests that keeping the R4 fate off in R3 precursors is as important as inducing the R3 fate.

R4 specification

Previous work has demonstrated that Fz/PCP-signaling leads to Dl and neur upregulation in R3, activating Notch-signaling in the neighboring R4 precursor (Cooper and Bray, 1999; del Alamo and Mlodzik, 2006; Fanto and Mlodzik, 1999; Tomlinson and Struhl, 1999). Here, we show that Egfr-signaling is also specifically required for R4 fate determination. The ETS factors Yan and Pnt are nuclear effectors of Egfr-signaling in many contexts including photoreceptor induction (Rebay and Rubin, 1995; Simon, 2000), and our data indicate that they act also in R3/R4 determination. Egfr-signaling leads to an inactivation of Yan and an activation of Pnt through their phosphorylation by the Rl/Erk MAPK. As Yan represses the R4 fate it needs to get inactivated in the R4 precursor by Egfr-signaling and conversely Pnt is activated in R4. Together with the Notch-Su(H) activity this leads to R4 fate induction. Thus, for R3 determination Fz/PCP-signaling and its nuclear effectors Fos (and Jun) are sufficient, along with Yan mediated repression of the R4 fate in R3 precursors. R4 fate determination, on the other hand, requires the joint activity of two pathways, Notch and Egfr-signaling and their nuclear effectors (see Fig. 7D). A similar Egfr-Notch cooperation is observed in R7 induction (Cooper and Bray, 2000; Tomlinson and Struhl, 2001) and in cone cells (Flores et al., 2000).

Our data support a complex interaction scenario between Fz/PCP, Notch, and Egfr-signaling in R3/R4 fate determination. Whereas the Notch-Su(H) activation in R4 depends on Fz/PCP-signaling in the R3 precursor, the Fz/PCP and Egfr-signaling pathways require a fine balance. This is reflected by their genetic interactions, both at the level of the receptors fz and Egfr (Fig. 6A,B) and their nuclear effectors, Fos/Jun and the ETS factors Pnt and Yan (Fig. 4B,C), suggesting a cooperative involvement between the Fz/PCP and Egfr-pathways.

The nuclear Egfr-signaling response is very likely mediated by Pnt in R4. Although this could not be addressed in pnt LOF clones due to the non-autonomous defects, which are caused by feed-back loop requirements in which Pnt participates (Casci and Freeman, 1999; Freeman, 1998). The sufficiency experiments fully support a cell-autonomous requirement of Pnt in R4 to specify R4 fate, consistent with the Egfr requirement (Fig. 7B).

In summary, the behavior of the nuclear effectors of the respective signaling pathways involved in R3/R4 specification, reflect the combinatorial nature of the signaling pathway input into the R3 and R4 fates.

Complex interactions of Fos and Jun in imaginal discs

Although in the embryo Fos and Jun need to act as heterodimeric partners in a non-redundant manner (Ciapponi et al., 2002), in imaginal discs the scenario is more complicated. Whereas jun mutant clones display only mild phenotypes and do not affect proliferation/survival (Weber et al., 2000), strong kay/fos LOF alleles show severe defects, suggesting that kay/fos is the main AP-1 component acting in imaginal discs. This is supported by recent studies on the role of Fos in cell cycle regulation and proliferation (Hyun et al., 2006). Nevertheless, the double mutant combination of kay and jun revealed a requirement of both, as no kay/fos, jun double mutant cells are recovered, suggesting a partially redundant function of kay/fos and jun in imaginal discs.

The specific role of the possible distinct heterodimers between the different Fos isoforms and Jun, or the different Fos isoforms themselves, could be very complex. This complexity is also evident in the fact that overexpression of a dominant-negative Fos protein form or a single wild-type isoform (transcript RA, according to Flybase) causes similar phenotypic defects (e.g. in the eye or in thorax closure). Future experiments will have to address which of the Fos isoforms is required in which context and if and how they interact with Jun.