Extramacrochaetae functions in dorsal-ventral patterning of Drosophila imaginal discs

Complex organs such as the mammalian retina are composed of several distinct functional units and hundreds, if not thousands, of different cell types. During development several mechanisms are put in place to ensure that each cell knows its exact geographical position so that it will adopt the appropriate fate for its location. A reliable piece of positional information is the relative location of each cell along three major axes: anterior-posterior, dorsal-ventral (D/V) and proximal-distal. Each axis serves as a signaling center that influences cell fate choices across the developing tissue by secreting short- and long-range morphogens. Cells that lie at varying distances from each axis must interpret the relative intensity of each signal and activate appropriate gene expression programs. Failure to specify each axis or to generate the polarizing signals that emanate from these singling centers can have catastrophic effects on organ development and function. Here, we use the Drosophila eye to examine the role that the helix-loop-helix (HLH) protein Extramacrochaetae (Emc) plays in patterning across the D/V axis.

The adult eye contains ∼750 unit eyes, or ommatidia, that are evenly placed within dorsal and ventral compartments. Each ommatidium contains eight photoreceptors (R1-8) that appear as an asymmetric trapezoid, with the R3 and R4 cells occupying unbalanced positions within the unit (Dietrich, 1909; Chen, 1929; Waddington and Perry, 1960). The ommatidia in the dorsal and ventral halves are mirror images of each other and where they meet in the adult retina is referred to as the equator (Ready et al., 1976). In order to generate these mirror-image arrangements, the photoreceptor clusters must first adopt different chiralities and then rotate in opposite directions. D/V patterning in the developing eye takes place during the larval stages and can be divided into early and late phases. In the early phase (late first/early second instar) the unpatterned eye is subdivided into dorsal, ventral, and midline compartments. In the latter phase (third instar), ommatidia adopt the correct chirality and rotate in the direction that is appropriate for their location in either the dorsal or ventral compartment.

In nearly all imaginal discs, the D/V axis is laid down during the late first/early second larval instar (Bohn, 1967; Bryant, 1970; Garcia-Bellido and Merriam, 1971a,b; Lawrence and Morata, 1976; Williams et al., 1993). Prior to the imposition of dorsal identity, the entire eye disc expresses the ventral selector gene fringe (fng). During the late first/early second instar, expression of pannier (pnr), which encodes a GATA transcription factor, is activated via an unknown mechanism in a small group of cells within the peripodial membrane along the dorsal margin (Heitzler et al., 1996; Maurel-Zaffran and Treisman, 2000; Oros et al., 2010). Pnr is then responsible for inducing the expression of wingless (wg) (Maurel-Zaffran and Treisman, 2000), which in turn activates expression of the Iroquois complex (Iro-C) genes within the dorsal half of the eye (McNeill et al., 1997; Heberlein et al., 1998). One member of Iro-C, mirror (mirr), represses the expression of fng in the dorsal half of the eye field (Cho and Choi, 1998; Cavodeassi et al., 1999; Yang et al., 1999; Sato and Tomlinson, 2007). In the ventral half of the eye, sloppy paired 1 (slp1) represses mirr, which in turn preserves fng expression and ventral identity (Sato and Tomlinson, 2007). The confrontation of fng⁻ (dorsal) and fng+ (ventral) tissue leads to the differential activation of Delta (Dl) and Serrate (Ser) within the two compartments and the activation of Notch (N) signaling at the D/V midline (Panin et al., 1997; Cho and Choi, 1998; Dominguez and de Celis, 1998; Papayannopoulos et al., 1998). Notch signaling at the midline is necessary for the growth of the eye field and for the expression of four-jointed (fj) (Zeidler et al., 1999; Chao et al., 2004; Reynolds-Kenneally and Mlodzik, 2005; Gutierrez-Aviño et al., 2009). Fj has been linked to planar cell polarity (PCP) and for Notch- and JAK/STAT-dependent growth (Zeidler et al., 1999; Gutierrez-Aviño et al., 2009). Thus, the early division of the eye into dorsal, ventral and midline zones is important for growth and for providing relevant positional information for later chirality and rotation decisions.

The second phase of D/V patterning occurs during the third larval instar as neuronal specification takes place behind the morphogenetic furrow. The early ommatidium is symmetrical with respect to the position of the presumptive R3 and R4 cells. The presumptive R3 lies on the equatorial side of the photoreceptor cluster, while the R4 lies on the polar side (Ready et al., 1976; Tomlinson and Ready, 1987). This symmetry is broken by the activity of several factors, including the Frizzled (Fz) receptor and the Notch signaling pathway. Fz levels are maintained at higher levels in the presumptive R3 cell as compared with the R4 precursor and are required to translate different levels of an as yet unidentified ligand into distinct gene expression programs (Zheng et al., 1995; Tomlinson and Struhl, 1999). Elevated Fz in the R3 precursor induces the production of Dl, which in turn activates the Notch pathway in the neighboring R4 precursor (Cooper and Bray, 1999; Fanto and Mlodzik, 1999). As a result, the R3 and R4 photoreceptors are now functionally distinct from each other. Ommatidia mutant for fz exhibit defects in chirality as well as in the direction and degree of rotation. In many fz mutant ommatidia, the R4 precursor adopts an R3 fate (Zheng et al., 1995). Likewise, in ommatidia with reduced Notch signaling, both precursors adopt the R3 fate, whereas hyperactivation of the pathway induces both cells to adopt the R4 fate (Cooper and Bray, 1999; Fanto and Mlodzik, 1999; Tomlinson and Struhl, 1999). Thus, maintaining appropriate levels of Fz and Notch activity in the R3/4 photoreceptor pair is crucial for the establishment of PCP in the eye.

Two additional factors that influence PCP are the transmembrane cadherins Fat (Ft) and Dachsous (Ds) (Mahoney et al., 1991; Clark et al., 1995; Ishikawa et al., 2008). These proteins affect PCP in part by maintaining higher Fz receptor levels in the presumptive R3 cell as compared with the R4 precursor (Yang et al., 2002). ft is expressed in a relatively uniform pattern throughout the eye disc, whereas ds is expressed in a gradient with highest levels at the poles. Mutations in the Fat/Ds system cause strong PCP defects throughout the entire disc (Yang et al., 2002; Rawls et al., 2002; Simon, 2004; Brittle et al., 2012; Thomas and Strutt, 2012; Sharma and McNeill, 2013; Ayukawa et al., 2014). By contrast, the eyes of homozygous loss-of-function fj mutant alleles are characterized by relatively weak PCP defects (Zeidler et al., 1999). Although mutant fj clones show somewhat stronger phenotypes, including non-autonomous effects (Zeidler et al., 1999), these are still relatively mild compared with disruptions of the Ft/Ds or Fz/Notch systems.

Once the R3/4 cells have adopted their individual fates, their positions in relation to each other will shift slightly, thus breaking the symmetry of the photoreceptor cluster. Ommatidia in the dorsal and ventral halves of the eye then rotate 90° in opposite directions, forming a mirror image across the equator. Rotation is a two-step process: there is an initial 45° rotation followed by a pause for 3-4 columns (∼6-8 h), and then a second 45° rotation (Choi and Benzer, 1994). Fz regulates the expression of one rotation gene, nemo, suggesting that there is a molecular connection between PCP and rotation (Zheng et al., 1995).

Emc encodes an HLH protein that binds and forms heterodimers with several basic HLH (bHLH) proteins, including Daughterless (Da) and members of the Achaete-Scute complex (AS-C) (Garrell and Modolell, 1990; Ellis et al., 1990; van Doren et al., 1991, 1992; Alifragis et al., 1997). Since Emc lacks the basic DNA-binding domain, neither Emc itself nor Emc-bHLH heterodimers can interact with DNA (van Doren et al., 1991, 1992). Therefore, Emc functions to sequester bHLH proteins away from their target genes. Retinas lacking emc suffer from several developmental defects, including disruptions in PCP and ommatidial rotation (Brown et al., 1995; Bhattacharya and Baker, 2009). The latter report demonstrated that, despite the loss of emc, the vast majority of R3/4 cells are specified correctly and that Notch signaling appears normal in the R4 cell. Based on these observations, it can be concluded that although emc is expressed in the R3/4 pair it is unlikely that the PCP and rotation defects seen in emc mutants are due to the loss of emc in these cells.

In a prior report, we have shown that emc expression is enriched along the D/V midline (Spratford and Kumar, 2013). This enrichment persists at the midline ahead of the advancing furrow even in late third instar discs (Brown et al., 1995; Bhattacharya and Baker, 2009; Spratford and Kumar, 2013). Here we demonstrate that Emc does not play a role in establishing the D/V midline but rather is expressed here in response to the formation of the midline. We also show that emc lies downstream of N signaling to regulate fj expression at the midline. Prior reports had demonstrated that Notch signaling regulates fj expression (Zeidler et al., 1999; Gutierrez-Aviño et al., 2009) and the implicit model has been that Notch signaling via Suppressor of Hairless [Su(H)] directly regulates fj expression. Our data support an alternative model in which Notch pathway activation of emc is required to maintain fj expression at the midline. Since relatively weak PCP phenotypes are seen in fj mutant retinas (Zeidler et al., 1999; Strutt et al., 2003), our data suggest that Emc influences PCP by also regulating other factors throughout the eye field. These factors are likely to lie outside of the Ft/Ds system, as ds expression is normal in emc loss-of-function clones. One potential target of Emc is likely to be the Wg pathway, as we have previously shown the Emc regulates wg transcription along the ventral margin (Spratford and Kumar, 2013). Our results indicate that Emc functions in both the early and late phases of D/V patterning.

In their seminal report on the role that Emc plays in eye development, Brown and colleagues reported that retinal clones of a hypomorphic emc mutant allele (emc¹) display slight PCP defects (Brown et al., 1995). A subsequent report using an emc null allele (emc^AP6) showed that ommatidia also fail to rotate correctly when the entire eye is mutant for emc (Bhattacharya and Baker, 2009). To study the mechanism by which Emc influences D/V patterning we first determined the expression pattern of emc in the eye, using a polyclonal antibody that recognizes Emc protein and six enhancer trap lines (four GFP and two lacZ insertions) that are located within 822 bp upstream of the transcriptional start site. We found that all six transcriptional reporters show expression throughout the entire disc as well as substantial enrichment along the D/V midline (Fig. 1A,B) (Spratford and Kumar, 2013). The transcriptional reporters are likely to provide an accurate view of emc expression since the antibody also reveals an enrichment of Emc protein at the midline (Fig. 1C).

The strikingly high midline expression that is revealed by the enhancer traps suggested that the GFP and lacZ reporters are under the control of a nearby D/V midline-specific enhancer element. In an attempt to identify this enhancer, we cloned 4 kb of DNA upstream of the transcriptional start site (surrounding the insertion sites of all six enhancer traps) but, unfortunately, several fragments containing portions of the upstream DNA sequence failed to drive expression of a lacZ reporter at the midline (supplementary material Fig. S1A-G). We also screened five GAL4 strains from the Janelia Farm Research Campus that contained DNA fragments from the emc locus and again failed to observe any midline expression (supplementary material Fig. S1A,H-Q). Therefore, the midline enhancer is likely to be located further away than we initially predicted.

We next set out to register the spatial and temporal expression pattern of emc with factors involved in D/V patterning and PCP. We first compared emc expression with that of mirr and slp1, which are expressed and function in the dorsal and ventral halves of the eye, respectively (McNeill et al., 1997; Sato and Tomlinson, 2007). The early eye disc lacks a sharp equator and instead has a broad midline region that is ∼10-15 cell diameters wide. Whereas neither mirr nor slp1 is expressed within this zone (Sato and Tomlinson, 2007), emc is highly enriched within these midline cells (Fig. 1D-I). In addition to emc, several genes that function in D/V patterning, such as Ser, Notch and fj, are also expressed within this zone (reviewed by Jenny, 2010). We were particularly interested in fj because PCP defects (albeit mild) in the eye are associated with loss-of-function mutant alleles (Zeidler et al., 1999; Strutt et al., 2003). We compared the expression of emc and fj transcriptional reporters and it appears that the enrichment of emc occurs within a large subset of cells that also express fj (Fig. 1J-L). The spatial location of emc at the midline and its overlapping expression with fj suggest a role for Emc in D/V patterning.

We sought to determine whether emc expression is dependent upon the establishment of the midline or if Emc itself is required to induce midline formation. To address the first model, we started by analyzing emc expression in mutants in which the midline has been abolished. Lobe (L¹) mutants are characterized by extensive cell death within the ventral half of the eye, thus leaving only dorsally fated tissue (Chern and Choi, 2002; Singh and Choi, 2003; Singh et al., 2006). In the L¹ mutant eye field, we find that although emc is expressed normally within the antenna and ahead of the advancing furrow, the enrichment of emc at the midline is entirely lost (Fig. 2A,B). Without the juxtaposition of dorsal and ventral tissue the midline itself probably fails to be specified.

We performed the reciprocal experiment and analyzed emc expression in animals with only ventrally fated tissue. This was accomplished by using DE-GAL4 (Morrison and Halder, 2010) to drive expression of the ventral selector genes fng and slp1 in the dorsal half of the eye. Similar to L¹ mutants, emc expression along the midline is lost (Fig. 2C-F). In these two cases it is also likely that the midline fails to be specified. Expression of ventral selector genes in the dorsal half of the eye field leads to small and disorganized adult eyes (Fig. 2G,H). The failure to produce a midline is expected to yield small eyes because the midline also serves as the site of Notch- and JAK/STAT-mediated eye growth (Bach et al., 2003; Chao et al., 2004; Tsai and Sun, 2004; Reynolds-Kenneally and Mlodzik, 2005). We also expressed fng and slp1 throughout the eye using an ey-GAL4 driver and in both cases observed that emc is no longer enriched at the midline (supplementary material Fig. S2A-F). Curiously, slp1-lacZ expression is still restricted to the ventral side of the eye disc despite the use of the ey-GAL4 driver. This might be due to the weak expression of the ey-GAL4 driver or its sometimes patchy and inconsistent expression pattern. We can conclude, however, that in all five genetic backgrounds in which we disrupted normal D/V patterning emc enrichment at the midline region of the eye disc is completely abolished (Fig. 2A-F; supplementary material Fig. S2A-F). These results suggest that the juxtaposition of dorsal and ventral tissue is required for emc to be expressed along the midline.

We then asked if the creation of artificial D/V boundaries could lead to the ectopic activation of emc expression. We used the flp-out system (Struhl and Basler, 1993; Pignoni et al., 1997) to generate clones that overexpress either mirr in the ventral compartment or fng in the dorsal compartment. These manipulations have been reported to result in the creation of new fng⁺/fng⁻ boundaries (Cho and Choi, 1998; Dominguez and de Celis, 1998; Papayannopoulos et al., 1998; Sato and Tomlinson, 2007). Echoing wild-type development, the formation of new fng⁺/fng⁻ boundaries leads to the loss of mirr, fng and slp1 and a subsequent elevation of Ser expression in a narrow band of cells surrounding the overexpression clones (Sato and Tomlinson, 2007). Since these new D/V boundaries have many hallmarks of the normal midline, we set out to determine whether an enrichment of emc expression accompanies the formation of these new boundaries. We were able to confirm this hypothesis as emc expression is indeed elevated in the cells bordering the mirr and fng overexpression clones (Fig. 3A-C,G-I, arrows).

Similar effects were observed when we overexpressed fng in the ventral compartment of the wing disc. Compared with the eye, the fng expression pattern in the wing is reversed, with fng being normally expressed solely within the dorsal compartment (Irvine and Wieschaus, 1994; Kim et al., 1995). When fng-expressing clones cross the endogenous D/V border and extend into the ventral compartment, enrichment of emc expression is shifted congruently with the formation of a new midline (Fig. 3D-F, arrows).

An additional case in which we can study the effect that the establishment of new fng⁺/fng⁻ boundaries has on emc expression is when mirr expression is removed from the dorsal compartment. fng expression is ectopically activated in iro^DFM3 null mutant clones (which are deficient for all three Iro-C genes) that lie within the dorsal half of the eye (Sato and Tomlinson, 2007). The generation of new D/V boundaries by the loss of the Iro-C genes in the dorsal compartment leads to a non-autonomous elevation of emc expression in cells surrounding the clone (Fig. 3J-L, arrow). Not all clones within the dorsal compartment show this non-autonomous enrichment within a single disc. We observe an elevation in emc expression surrounding the majority of clones and also note that there does not appear to be a domain restriction for this effect – we readily see emc upregulation surrounding clones that contact the margin or lie within either the anterior or the posterior compartments. Altogether, we conclude that emc expression at the midline, in both the eye and wing, is induced by the juxtaposition of dorsal and ventral compartments.

We next determined whether Emc participates in the establishment of the midline. We generated flp-out overexpression clones of emc and analyzed the expression of several genes, including mirr, slp1 and fj. As the expression of these genes remains unaltered in emc overexpression clones (Fig. 4A-L), we conclude that Emc cannot induce the formation of a D/V boundary. Since Emc loss-of-function mutants have chirality and rotation defects we also examined whether the overexpression of emc could induce PCP and rotation defects. We used a transcriptional reporter for spalt major (salm-lacZ), which is expressed in the R3/4 pair, to visualize rotating photoreceptor clusters (Barrio et al., 1999; Fanto and Mlodzik, 1999; Domingos et al., 2004). Overexpression of emc failed to modify the chevron-like pattern of salm expression (compare Fig. 4M-P with Fig. 7A,E), suggesting that increased levels of emc cannot alter ommatidial chirality and/or rotation.

The Notch pathway regulates emc expression in several developmental contexts, including vein and margin formation in the wing disc, patterning mesodermal segments within the developing embryo, and neural determination in the eye (Baonza et al., 2000; Baonza and Freeman, 2001; Tapanes-Castillo and Baylies, 2004; Bhattacharya and Baker, 2009). In the first two instances, Notch activates the expression of emc within the wing and embryo. However, in the developing eye Notch appears to either promote or repress emc depending upon the developmental context. Within the morphogenetic furrow atonal (ato) expression appears to depend upon the repression of emc by the Notch pathway (Baonza and Freeman, 2001). By contrast, Notch activation of emc transcription is required for R7 and cone cell development (Bhattacharya and Baker, 2009). We sought to determine if Notch signaling regulates emc expression at the D/V midline and, if so, whether it is via activation or repression.

We first increased Notch signaling by expression of its intracellular domain (N^icd) in flp-out clones and observed that this manipulation is sufficient to activate an emc-lacZ reporter cell-autonomously (supplementary material Fig. S3A-D, arrow). To assess the necessity of the Notch pathway in maintaining emc expression we used several different genetic manipulations to reduce the level of Notch activity. First, we used a temperature-sensitive allele of Notch (N^ts1; Shellenbarger and Mohler, 1975) to block Notch pathway activity and observed that emc expression at the midline is specifically eliminated (compare Fig. 1A-C with Fig. 5A-C, arrows), whereas emc expression throughout the rest of the eye field appears to be unaffected by the loss of Notch activity. Second, we expressed a dominant-negative allele of Ser (Ser^DN; Hukriede et al., 1997) as an alternate approach to reducing Notch activity. In flp-out clones expressing Ser^DN we also observed downregulation of emc expression at the midline (Fig. 5D-F, arrow). Third, we removed Notch activity by inducing clones of a Su(H) loss-of-function mutant allele and again observed that emc expression along the midline within the eye and wing is lost (Fig. 5G-M, arrow). Lastly, we generated flp-out overexpression clones of a dominant-negative form of mastermind (mam^DN), which encodes a co-factor for Su(H) (Helms et al., 1999). In this case, as with all the others, emc expression is lost when mam^DN is expressed either in clones (Fig. 5N-P, arrow) or throughout the entire eye field (Fig. 5Q-S). Su(H) and Mam are DNA-binding proteins and therefore the pattern of emc loss is informative: reduction in both proteins leads to the cell-autonomous loss of emc expression and this supports the contention that the Notch pathway directly regulates emc transcription. These findings are also consistent with the cell-autonomous activation of emc in N^icd overexpression clones (supplementary material Fig. S3A-D). The Notch pathway is required not only at the midline, as clones that include non-midline tissue in the wing also show reductions in emc expression (Fig. 5J-M).

The Notch pathway bifurcates downstream of Su(H), with genes of the Enhancer of split complex [E(spl)-C] residing within one branch (de Celis et al., 1996). The forced expression of several E(spl)-C genes (mβ, m5 and m8) does not affect emc expression (supplementary material Fig. S4A-L), suggesting that emc lies within a separate branch of the cascade. Overall, our results support a model in which an E(Spl)-independent branch of the Notch pathway directly activates emc expression at the D/V boundary within developing eye and wing discs.

fj is expressed in a gradient within the eye disc, with the highest levels at the midline (Fig. 1J) (Brodsky and Steller, 1996; Strutt et al., 2003). Similarly, in the wing disc fj is expressed at the highest level within central regions of the pouch and at lower levels in peripheral regions (Fig. 6D) (Strutt et al., 2003). The Notch pathway has been proposed to directly regulate fj expression in both tissues (Zeidler et al., 1999). However, the conspicuous overlap of fj and emc expression at the midline in the eye and wing, as well as the requirement for Notch signaling in maintaining emc expression, prompted us to test a model in which Notch regulation of fj requires Emc activity. We generated emc^AP6 null clones in the eye and wing and analyzed the expression of a fj-lacZ reporter. In both tissues, we observe a cell-autonomous reduction in fj expression (Fig. 6A-C,F,G, arrow). fj-lacZ expression is not completely eliminated in emc mutant tissue, which is consistent with previous findings that additional signals from the JAK/STAT and Wg pathways also regulate fj expression at the midline (Zeidler et al., 1999; Gutierrez-Aviño et al., 2009). To determine if JAK/STAT regulation of fj is also through Emc we analyzed emc expression in both Stat92E overexpression (supplementary material Fig. S5A-C) and loss-of-function (supplementary material Fig. S5D-F) clones. In both cases, we see no effect on emc expression and therefore conclude that Notch activates fj expression via emc, while JAK/STAT signaling has a separate input into fj.

The PCP defects seen in clones of the emc¹ hypomorphic allele are very mild (Brown et al., 1995) and probably do not reveal the full role that Emc plays in either ommatidial rotation or PCP. We generated and analyzed both types of defects in emc^AP6 null mutant clones. On their own, emc^AP6 mutant cells proliferate very slowly and thus the clones are very small and difficult to analyze. To overcome this obstacle we provided the emc^AP6 cells with a growth advantage by generating the clones in a Minute (M3ⁱ⁵⁵) heterozygous mutant background. When placed against M(3)ⁱ⁵⁵/+ heterozygous tissue the emc^AP6 null mutant clones are large and can be analyzed for both rotation and PCP defects in the imaginal disc and in adult retinal sections. In emc^AP6 clones, the majority of ommatidia have properly specified R3/4 pairs but the chevron-like arrangement is substantially disrupted (Fig. 7A-F), suggesting that ommatidial rotation has not been completed properly. In Fig. 7F it is possible to see ommatidia that have not rotated correctly (green arrows).

In order to determine whether chirality is also affected, we looked at the orientation of ommatidia in adult retinal sections. In a normally constructed ommatidium, the rhabdomeres of the R1-7 photoreceptors are arranged in an asymmetrical trapezoid (Fig. 7G) (Dietrich, 1909; Chen, 1929; Waddington and Perry, 1960). The trapezoids of the dorsal and ventral compartments are arranged as mirror images of each other with the two chiral forms meeting at the equator (Ready et al., 1976). Within emc null clones a significant number of ommatidia have either fewer than the normal number of photoreceptors (Fig. 7H,I, red circles) or have undergone significant degeneration (Fig. 7H,I, blue circles). In both cases we were unable to determine the chirality of the ommatidia. However, in several instances we were able to clearly see ommatidia with inappropriate chirality (Fig. 7H,I, purple arrow) or ommatidia in which both the R3 and R4 precursor cells have adopted the R3 fate (Fig. 7H,I, red arrows). By far the most common defect in the adult retina is the incorrect rotation of ommatidia (Fig. 7H,I, green arrows), which also seemed to be the predominant phenotype in the developing eye disc (Fig. 7F). Since so many ommatidia could not be scored for either chirality or rotation, our analysis most likely underestimates the role that Emc plays in PCP and rotation.

As fj loss-of-function mutants have mild chirality defects and no rotation defects we investigated whether emc regulates the Ft/Ds system. In emc loss-of-function clones, ds expression levels (using a ds-lacZ reporter) appear relatively normal (supplementary material Fig. S6A-H), suggesting that Emc influences PCP by regulating other factors outside of the Ft/Ds system. Since wg expression is reduced in emc clones that contact the margin (Spratford and Kumar, 2013), it is possible that the PCP and rotation defects that are seen in emc mutant tissue are in part due to the regulation of Wg signaling.

Two prior studies have indicated that Emc is a key player in patterning the D/V axis of the developing Drosophila eye. Both reports provide evidence that ommatidia lacking emc have a propensity to adopt the incorrect chiral form and fail to rotate properly (Brown et al., 1995; Bhattacharya and Baker, 2009). However, the mechanism by which Emc exerts controls over D/V patterning has not been elucidated. The latter report did suggest that Emc does not function within the R3/4 photoreceptor pair to promote PCP. These cells use the Fz and Notch pathways to receive and amplify a polarizing signal (Zheng et al., 1995; Cooper and Bray, 1999; Fanto and Mlodzik, 1999; Tomlinson and Struhl, 1999). In emc mutant tissue the vast majority of R4 cells are properly specified and Notch signaling, as assayed by an E(spl)mδ reporter, appears normal (Bhattacharya and Baker, 2009). Thus, it is unlikely that emc, although expressed in the R3/4 neurons, functions in these cells to promote ommatidial chirality and rotation (Bhattacharya and Baker, 2009).

The other sites within the developing eye at which Emc could function to promote PCP are the margins where Wg and Ds levels are highest and at the D/V midline where fj expression is enriched (Fig. 8A,B). We have previously shown that in young eye discs emc expression is distinctly enriched at the midline (Spratford and Kumar, 2013). This enrichment persists in the anterior compartment in older discs (Brown et al., 1995; Bhattacharya and Baker, 2009; Spratford and Kumar, 2013); therefore, Emc may regulate fj expression at the midline. In that same study we also demonstrated that Wg expression at the ventral margin depends on emc expression (Fig. 8B) (Spratford and Kumar, 2013) and therefore Emc could, in principle, also regulate Ds protein levels.

In this study we set out to determine the mechanism by which Emc functions in D/V patterning. We first refined our analysis of emc transcription and showed that emc is enriched within a zone of cells at the midline that lies between the expression domains of the dorsal selector gene mirr and the ventral compartment gene slp1 (Fig. 1A-I). The expression of emc at the midline appears to be dependent upon the juxtaposition of dorsal and ventral fated tissue. When we generate an entire field made up of a single compartment (either dorsal or ventral), enrichment of midline emc expression is lost (Fig. 2A-F; supplementary material Fig. S2A-F), and if we induce new D/V boundaries emc expression is ectopically induced at the new midline (Fig. 3J-L). It appears that Emc does not play a role in establishing the midline, as ectopic expression of emc is insufficient to induce ectopic D/V borders (Fig. 4A-P).

How is emc expression activated at the midline in response to the juxtaposition of fng⁺ and fng⁻ cells? The confrontation of these two populations of cells leads to the activation of Notch signaling at the midline. Emc has been placed downstream of the Notch pathway in several developmental contexts, including developing photoreceptors and the wing. Here, we have determined that Notch signaling promotes emc expression at the midline via the Su(H)-Mam complex (Fig. 5, Fig. 8A,B; supplementary material Fig. S3A-D). What is the function of Emc at the midline? Prior reports had suggested that N signaling activates fj expression directly via Su(H) (Zeidler et al., 1999; Gutierrez-Aviño et al., 2009). Since emc expression overlaps significantly with fj at the midline (Fig. 1J-L), we tested the validity of an alternative model in which Notch activation of emc is required to regulate fj expression. We provide evidence that emc functions to maintain fj expression at the midline – fj expression is greatly reduced in emc loss-of-function clones (Fig. 6A-G, Fig. 8A,B). Also, as expression of emc is unable to induce fj expression it is more likely that Emc is instead required to maintain fj expression. Since fj plays a minor role in PCP within the eye, only a fraction of the chirality defects seen in emc mutant tissue are likely to be caused by the loss of fj. The majority of chirality defects and all of the rotation defects that are seen in emc mutants (Fig. 7) (Brown et al., 1995; Bhattacharya and Baker, 2009) are likely to be due to dysregulation of Wg signaling and other factors.

Four emc homologs encoded within the vertebrate genome are referred to as the inhibitor of DNA binding genes (Id1-4) (Benezra et al., 1990; Christy et al., 1991; Sun et al., 1991; Biggs et al., 1992; Ellmeier et al., 1992; Deed et al., 1993; Riechmann et al., 1994; Zhu et al., 1995). To date, a role for the Id proteins in D/V patterning of the vertebrate retina has not been reported. This might be a result of functional redundancy. All Id proteins are expressed within the retina and several are transcribed early in development. For example, within the mouse retina Id1-3 are expressed early in the optic vesicle and the overlying surface ectoderm, then later in the ganglion and neuroblastic layers of the optic cup, with expression persisting in the retinal ganglion cells and in both rod and cone photoreceptors (Ellmeier and Weith, 1995; Jen et al., 1996, 1997; Yeung and Yip, 2005; Du and Yip, 2011). By contrast, Id4 expression is restricted to the neural layer of the optic cup (Jen et al., 1996). It has been proposed that, like the fly retina, the vertebrate eye might also be born with just ventral identity and that dorsal fates are imposed later in development (Murali et al., 2005). This model is of particular interest, as one or more Id proteins might play a role similar to that of Emc in regulating D/V patterning in the vertebrate retina. It is intriguing to think that roles for Emc/Id in D/V patterning might not be limited to the retina. Manipulations of vertebrate PCP genes lead to developmental defects within the mammalian kidney (reviewed by Carroll and Yu, 2012). Id1-3 are expressed within the developing mouse kidney and in rat cultured kidney glomerular mesangial cells (Jen et al., 1996). If Emc/Id proteins are shown to play a conserved role in D/V patterning then considerable light might soon be shed on mammalian development and disease.

(1) emc-GFP^YB0067/TM3; (2) P{PZ}emc⁰⁴³²²/TM3; (3) w¹¹¹⁸; (4) w*; mirr^B1-12/TM6B; (5) P{PZ}slp1⁰⁵⁹⁶⁵/CyO; (6) fj-lacZI; (7) L¹; (8) eyFLP; DE-GAL4/TM6B; (9) UAS-fng, emc-GFP^YB0067/TM3; (10) y¹ w*; UAS-slp1/CyO; (11) hsFLP²²; (12) Act5C>y+>Gal4, UAS-RFP/TM3; (13) Act5C>y+>GAL4, UAS-GFP; (14) UAS-mirr; (15) y¹ w* eyFLP; (16) y¹ w* hsFLP; iro^DFM3 FRT2A/TM3; (17) Ubi-GFP FRT80B; (18) y¹ N^ts1 g² f¹/C(1)DX y¹ f¹; (19) UAS-Ser^DN, emc-GFP^YB0067/TM6; (20) Su(H)^Δ47 FRT40A [w+l(2)35Bg+]/CyO; (21) w¹¹¹⁸; Ubi-GFPnls FRT40A/CyO; (22) UAS-mam^DN; (23) y¹ w* hsFLP²²; emc^AP6 FRT80B/TM6B; (24) y¹ w* hsFLP²²; Ubi-GFP M(3)ⁱ⁵⁵ FRT80B/TM6B; (25) UAS-emc4M; (26) P{PZ}salm⁰³⁶⁰²/CyO; (27) emc-E1-lacZ; (28) emc-E2-lacZ; (29) emc-E3-lacZ/TM3; (30) GMR10D04-GAL4; (31) GMR10H11-GAL4; (32) GMR10B05-GAL4; (33) GMR10C04-GAL4; (34) GMR10B08-GAL4; (35) UAS-lacZ; (36) UAS-Stat92E; (37) w*; FRT82B Stat92E^85c9/TM6B; (38) FRT82B Ubi-GFPnls; (39) y¹ w* M{vas-int.Dm}ZH-2A; PBac{y+-attP-3B}VK00033; (40) y¹ v¹; P{TRiP.JF02100} attP2/TM3; (41) y¹ w*; UAS-E(spl)mβ^48.1; (42) UAS-E(spl)m8; (43) UAS-E(spl)m5; (44) UAS-N^icd; (45) salm-lacZ; (46) ey-GAL4. Stocks 1-7, 11-18, 20, 21, 26, 30-35, 38-43, 45 and 46 were from the Bloomington Drosophila Stock Center; stock 8 is from Georg Halder (VIB Vesalius Research Center, KU Leuven, Belgium); stock 9 is a combination of two stocks from the Bloomington Drosophila Stock Center; stock 10 is from Andrew Tomlinson (Columbia University, NY, USA); stock 19 is a combination of UAS-Ser^DN from Robert Fleming (Trinity College, Hartford, CN, USA) and emc-GFP^YB0067 from the Bloomington Drosophila Stock Center; stock 22 is from Barry Yedvobnick (Emory University, Atlanta, GA, USA); stocks 23 and 24 are from Nick Baker (Albert Einstein College of Medicine, NY, USA; stock 25 is from Antonio Baonza (Centro de Biología Molecular Severo Ochoa (CSIC/UAM), Madrid, Spain); stocks 27-29 were generated by C.M.S.; stocks 36 and 37 are from Erika Bach (New York University, USA); stock 44 is from Kevin Moses (Wellcome Trust, London, UK).

Primary antibodies: rat anti-Elav (1:100; Developmental Studies Hybridoma Bank); rabbit anti-Emc (1:1000; Yuh Nung Jan, UCSF, CA, USA); and mouse anti-β-gal (1:100; Promega, Z3781). Flurophore-conjugated secondary antibodies and phalloidin-flurophore conjugates were obtained from Jackson Laboratories and Molecular Probes. Imaginal discs were prepared as described by Anderson et al. (2012). Immunostaining was carried out according to Spratford and Kumar (2014). For scanning electron microscopy, adults or pharate adults were serially incubated in 25% ethanol, 50% ethanol, 75% ethanol, 100% ethanol, 50% ethanol:50% hexamethyldisilazane (HMDS) and 100% HMDS, coated with gold-palladium and viewed with a JEOL 5800LV scanning electron microscope. Retinal sections were prepared as described by Jenny (2010). Slides were photographed on a Zeiss Axioplan II compound microscope.

For Notch temperature-sensitive mutants, an emc-lacZ reporter was crossed to the N^ts1 strain and flies were allowed to lay eggs in vials at 25°C for 48 h. Adults were transferred to new vials, following which the old vial was incubated at 16°C for 24-96 h. Once first instar larvae were visible, vials were moved to the restrictive temperature of 30°C for 24-48 h before dissection of early third instar larvae. All heat shock-induced clones were induced at 37°C 48 h after egg lay. emc^AP6 loss-of-function clones were induced by a 20 min heat pulse followed by a 60 min rest period (at room temperature) and a second 20 min heat pulse. Overexpression and MARCM clones were induced with 20 min and 60 min heat pulses, respectively. At least ten discs containing clones were analyzed for each genotype. Each phenotype documented within figures was observed in at least 50% of clones.

The three putative enhancer regions illustrated in supplementary material Fig. S1 were amplified from w¹¹¹⁸ genomic DNA by PCR and cloned into the placZ.attB plasmid (Konrad Basler, University of Zurich, Switzerland). Cloning strategies and PCR conditions are described in the supplementary Methods; for primers see supplementary material Table S1. The p.enhancer.lacZ.attB constructs were transformed into flies and stable stocks were created and analyzed for lacZ expression in imaginal and brain tissue.

We thank Brandon Weasner and Bonnie Weasner for technical support with the cloning of enhancer constructs and the generation of transgenic fly strains; Rudi Turner for generating retinal sections and scanning electron micrographs of adult eyes; Konrad Basler, Andrew Tomlinson, Yuh-Nung Jan, Nick Baker, Alan Spradling, Lynn Cooley, Antonio Baonza Erika Bach, Robert Fleming, Georg Halder, Kevin Moses, Barry Yedvobnick and the Bloomington Drosophila Stock Center for DNA plasmids, antibodies and fly stocks; and Utpal Banerjee for generously allowing C.M.S. to complete a portion of this work in his lab.