The Id protein Extramacrochaetae restrains the E protein Daughterless to regulate Notch, Rap1, and Sevenless within the R7 equivalence group of the Drosophila eye

Transcription factors of the helix-loop-helix (HLH) class play important roles in specification of cell fates. Multiple proneural bHLH genes act as master regulators of neural fate specification and differentiation in metazoans, including Drosophila and mammals (Baker and Brown, 2018; Dennis et al., 2019; Johnson, 2020). The regulation of these transcription factors helps control the timing and pattern of neural development. Proneural bHLH proteins require ubiquitous bHLH proteins called E proteins as heterodimer partners. They can also heterodimerize with Inhibitor of DNA binding (Id) proteins, HLH proteins that lack basic sequences, preventing DNA binding by proneural proteins and hence preventing transcription factor function. Accordingly, E proteins and Id proteins are believed to define competence for neuronal specification and differentiation in response to proneural proteins (Bertrand et al., 2002; Ling et al., 2014; Oproescu et al., 2021; Roschger and Cabrele, 2017; Troost et al., 2015; Wang and Baker, 2015a).

Both the proneural bHLH genes of the Achaete-Scute gene Complex (AS-C), and the unlinked proneural gene atonal (ato), are important during Drosophila eye development (Cadigan et al., 2002; Jarman et al., 1994). Each Drosophila compound eye is composed of ∼800 individual facets, called ommatidia, which consist of eight photoreceptor neurons (R1-R8), four cone cells and some other accessory cells. Cell fate specification begins during the third larval instar with R8 photoreceptor cell in the morphogenetic furrow, a visible indentation that progresses anteriorly across the eye imaginal disc until the whole eye field is differentiating (Fig. 1A,B) (Ready et al., 1976). R8 specification depends upon the proneural gene ato (Jarman et al., 1994). After R8 has been specified, short range signals from R8 cells recruit photoreceptor cells R2, 3,4,5. Further signals from the resulting five-cell preclusters then specify R1/6/7 cells, and finally non-neuronal cone cells in a still further round of recruitment (Freeman, 1997; Tomlinson et al., 1987; Treisman, 2013). After pupariation, sensory neural structures called inter-ommatidial bristles form in between ommatidia (Cagan and Ready, 1989; Ready et al., 1976). Inter-ommatidial bristles depend on AS-C proneural genes (Cadigan et al., 2002).

All the Ato and AS-C proteins require Daughterless (Da), their ubiquitously expressed bHLH protein heterodimer partner, for DNA binding and transcriptional activity. Da is also required, without Ato or AS-C, for the differentiation of photoreceptor cells R2-R5, and Da protein levels are elevated within the morphogenetic furrow where R8 and R2-5 cell fates are specified (Fig. 1A-C) (Brown et al., 1996). The only Id protein in Drosophila is encoded by the extramacrochaetae gene (emc) and is also widely expressed, but levels are reduced within the morphogenetic furrow (Fig. 1D) (Brown et al., 1995). Loss of emc function in the eye accelerates morphogenetic furrow progression and overall eye differentiation (Bhattacharya and Baker, 2012; Brown et al., 1995).

R1,6,7 and cone cell specification occur independently of both Da and of ato or AS-C (Brown et al., 1996; Jarman et al., 1994; Jiménez and Campos-Ortega, 1987). These cells are specified from an equivalence group of precursors by cell interactions. R7 specification requires the combinatorial activity of Notch and of the receptor tyrosine kinase Sevenless (Sev) (Fig. 1E) (Cooper and Bray, 2000; Mavromatakis and Tomlinson, 2012a; Miller et al., 2009; Tomlinson and Struhl, 2001). Expression of the Notch ligand Delta (Dl) on R1 and R6 cells is thought to activate Notch in the R7 precursor, and inhibit it in R1 and R6 precursors, while expression of Bride of Sevenless (Boss) on R8 cells activates signaling by its receptor Sevenless in R7 (Cooper and Bray, 2000; Mavromatakis and Tomlinson, 2012a; Miller et al., 2009; Tomlinson and Struhl, 2001; Van Vactor et al., 1991). The other cells within the equivalence group receive Dl but not Boss, and Notch signaling leads these cells to cone cell fates (Fig. 1E) (Flores et al., 2000). In sevenless mutants, the presumptive R7 cell is transformed to the cone cell fate by N activity (Mavromatakis and Tomlinson, 2012a; Tomlinson and Ready, 1986). Conversely, ectopic Sev signaling converts presumptive cone cells to R7 photoreceptor fate (Dickson et al., 1992; Treisman, 2013; Van Vactor et al., 1991).

R7 cell fate also requires activity of Rap1, a small GTPase proposed to synergize with Ras (Mavromatakis and Tomlinson, 2012b,a). Rap1 maintains the apical localization of Sevenless protein, as well as expression of adherens junction proteins (Baril et al., 2014), consistent with the requirement for Rap1 in maintaining cell adhesion by undifferentiated Drosophila imaginal disc cells (Knox and Brown, 2002). Rap1 also remodels adherens junctions during wound healing (Rothenberg et al., 2023). In vertebrates, Rap1 is required for integrity of retinal tissue in the optic tectum (Maddala et al., 2015). Rap1 activity is promoted by Guanosine Nucleotide Exchange Factors, and inhibited by GTPase activating proteins (GAPs) that convert Rap1-GTP into Rap1-GDP (Bos et al., 2007).

Surprisingly, both R7 and cone cell fates depend on emc, even though neither the proneural proteins nor Da are required. Loss of emc almost abolishes differentiation of R7 photoreceptors, as well as reducing and delaying differentiation of cone cells (Bhattacharya and Baker, 2009). Such requirements for Id proteins by cell types that do not require E proteins or other bHLH proteins have been noted before and raise the question of whether Id proteins also act through some non-bHLH-dependent mechanism (Ling et al., 2014; Wang and Baker, 2015a; Roschger and Cabrele, 2017; Oproescu et al., 2021).

One way that emc can act independently of proneural bHLH genes is by regulating da. This turns out to be why emc is required for growth of undifferentiated imaginal disc cells (Alonso and García-Bellido, 1988; Bhattacharya and Baker, 2011). Da protein levels increase when emc is mutated, due to transcriptional autoregulation and to Da protein stability in homodimers (Bhattacharya and Baker, 2011; Li and Baker, 2019). da, in turn, is required for Emc expression, through transcriptional regulation and increased Emc protein stability in heterodimers (Bhattacharya and Baker, 2011; Li and Baker, 2018). It is elevated Da that gives undifferentiated cells a competitive disadvantage during growth, mimicking emc mutations (Bhattacharya and Baker, 2011). Transcriptional targets of Da that regulate imaginal disc growth include the Hippo pathway gene expanded (Wang and Baker, 2015b), and potentially the cell cycle gene string (Andrade-Zapata and Baonza, 2014).

Here, we investigate whether Emc also restrains Da in eye development. While we reported previously that Da over-expression had no effect on R7 or cone cell development (Bhattacharya and Baker, 2009), more recent data question the interpretation of this finding. Because the Emc protein is made in excess but only stable in heterodimers with other HLH proteins such as Da, increasing Da expression increases levels of Emc/Da heterodimers by stabilizing more Emc, and the increment in Da homodimers, if any, is likely to be small (Li and Baker, 2018). We re-examined the role of emc in R7 and cone cell development here. We found that Emc indeed restrained Da activity during R7 photoreceptor and cone cell specification. We found evidence that Da was responsible for reduced Notch activity within the R7 equivalence group, and that emc also affected Sev and Rap1, the other signals controlling R7 specification. Aspects of the phenotype were mimicked by expressing Da in tethered homodimers that cannot heterodimerize with Emc. Thus, R7 and cone cell differentiation represent further examples where proneural-independent functions of emc reflect Da deregulation. This may be a common explanation for how emc, and potentially mammalian Id genes, affect developmental processes that are independent of proneural genes.

To test definitively whether emc restrains Da activity to permit R7 specification, we compared R7 development in da emc double-mutant clones and emc single-mutant clones. If emc is important in R7 because it restrains Da activity, then emc should not be required in the absence of the da gene, and R7 differentiation should be restored to da emc double-mutant cells. It is important in this regard that previous studies show that da is dispensable for the development of R1, R6 and R7 cells (Brown et al., 1996), Thus, R7 cells should be able to develop in da emc clones if emc is only required to regulate da. If emc is required in R7 for other reasons, R7 cells will not differentiate in da emc clones. Fig. 2 shows emc, da emc, and da clones in eye imaginal discs from late third instars. Their comparison is complicated by their different sizes and overall differentiation status. Whereas neuronal differentiation occurs throughout emc clones, albeit accelerated and abnormal in pattern (Bhattacharya and Baker, 2009, 2011; Brown et al., 1995) (Fig. 2A), da is required cell-autonomously for the differentiation of R8, R2, R3, R4 and R5 cells (Brown et al., 1996). Because R8 and perhaps others among these cells are required to recruit R1,6,7, and cone cells, central regions of da clones and da emc clones lack ommatidial differentiation due to the combined cell-autonomous and non-autonomous effects on the various ommatidial cell types (Fig. 2B,C). Cell-autonomous roles of emc and da in R7 can still be assessed in genetic mosaics, where R8 and R2, R3, R4 and R5 cells can be wild type for da while other cells are not. Such mosaic ommatidia occur at the boundaries of da and da emc clones (Fig. 2B,C).

The lack of differentiation within da and da emc clones leads to a further difference from emc clones. Since cell survival in the eye disc posterior to the furrow depends on signals emanating from differentiating ommatidia (Baker, 2001), the centers of da clones lack survival signals. Only small da clones were found far posterior to the furrow, reflecting rescue of da mutant cells near the clone boundaries by survival signals coming from wild-type cells and from differentiating R1, R6 and R7 cells (Fig. 2B). The da emc clones shared this same phenotype; only small da emc clones were found posterior to the furrow (Fig. 2C).

In emc mutant clones, cells occupying the R7 position in the ommatidium are not R7-like; they continue to express the pan-neuronal marker Elav, but fail to express at least four R7 cell markers [Runt, Prospero, Spalt, E(spl)mδ], and instead express Seven-up (Svp), a marker of R1/6 cells (Bhattacharya and Baker, 2009). Here, we used loss of Runt or Pros to follow the transformation of the R7 cell in emc mutants (Fig. 3). Runt is normally expressed by wild-type R7 cells from column 8 or 9 onwards, as well as by R8 cells (Kaminker et al., 2002). In the mosaic ommatidia that survive in da mutant clones, cells in the R7 position expressed both Elav and Runt (Fig. 3B; 40/41 cases), resembling normal R7 development. In da emc double mutant mosaic ommatidia that developed at the edge of da emc clones, cells in the R7 position also expressed Runt, which is typical of R7 cells (Fig. 3C; 25/25 cases). Therefore, da was epistatic to emc for these aspects of R7 development, indicating that elevated Da affects R7 development in emc mutant clones. Prospero is expressed in R7 cells from column 7 or 8 onwards (Kauffmann et al., 1996). Cells in the R7 position in emc mutant clones lack Pros expression (Fig. 3D) (Bhattacharya and Baker, 2009). By contrast, in both da and da emc clones, cells in the R7 position expressed both Elav and Prospero, confirming their R7-like differentiation with this second marker (40/41 for da, 25/25 for da emc; Fig. 3E,F). We can not exclude that examination of further markers, such as Svp, might reveal that cells in da emc clones occupy an intermediate state with some R7 and some R1/6-like properties. Taken together, however, these results do indicate that the failure of R7 development in emc clones depends on a cell-autonomous consequence of Da activity. There was no evidence for a requirement for emc in the absence of da.

In addition to R7 photoreceptor fate specification, emc is also required for the specification of non-neuronal cone cells. Only about half as many cone cells are specified in emc mutant ommatidia, and cone cells that remain delay expression of the homeobox transcription factor Cut by 2 to 3 columns and of Pros by 3 to 4 columns (Bhattacharya and Baker, 2009). To test whether Emc regulates cone cell specification through Da activity, cone cell development in mosaic ommatidia simultaneously mutant for both emc and da was compared with ommatidia mosaic for either emc or da alone (Fig. 4). As for R7 differentiation, cone cells were only recovered in mosaic ommatidia at boundaries of da or da emc clones. Cone cells expressing Cut or Pros but not Elav were recovered in da mutant mosaic ommatidia (Fig. 4A,C). The da emc double mutant mosaic ommatidia developed cone cells expressing Cut or Pros at similar rates (Fig. 4B,D). Although it is difficult to compare the frequency of cone cell differentiation to that in emc clones because only mosaic ommatidia at clone boundaries can be recovered for the da and da emc genotypes, there was no evidence that cone cell differentiation was delayed in da mutant cells, or that the timing of Cut or Pros expression was delayed in da emc cone cells compared to da cone cells. These results suggested that emc was not required for the rate or timing of cone cell differentiation in the absence of da, consistent with the notion that elevated Da is the cause of cone cell differentiation defects in emc clones.

The E(spl)-C includes seven genes that encode bHLH proteins of a distinct class that are transcribed in response to Notch signaling and function as transcriptional repressors that antagonize neurogenesis (Bray, 2006; Delidakis and Artavanis-Tsakonas, 1992; Jennings et al., 1994). A monoclonal antibody that detects four of the seven E(spl) bHLH proteins detects expression in many eye disc cells posterior to the morphogenetic furrow (Baker et al., 1996; Dokucu et al., 1996; Jennings et al., 1994). All the expression in differentiating R cells depends on Notch, although this is not the case for expression in undifferentiated retinal precursor cells (Baker and Yu, 1997). Previously, we reported that E(spl)-C protein expression in the R1, R6 and R7 cells was delayed in emc mutants (Bhattacharya and Baker, 2009) (Fig. 5A,B). Thus, emc contributes to Notch signaling in R1, R6 and R7 cells. To address whether it was Da that inhibits Notch signaling in emc mutants, E(spl) antibody was used to label da emc clones. In contrast to emc mutant clones (Fig. 5B), E(spl) expression was restored to da emc mutant R1, R6 and R7 cells, and not delayed (Fig. 5C-G). This was consistent with Da being the cause of reduced Notch signaling in emc mutants. We showed previously that emc mutations affected a transcriptional reporter of Notch signaling as well as E(spl) proteins (Bhattacharya and Baker, 2009). Because we studied E(spl) protein in da emc clones, we cannot distinguish whether Notch signaling is restored at the level of N signal transduction to the nucleus, or by direct Da-E(spl) protein interactions and changes in E(spl) protein stability in the absence of Da (Zarifi et al., 2012) (Kiparaki et al., 2015).

It was surprising that da was required for R7 and cone cell fates, because we previously reported that da overexpression had no effect on these cell types. More recent studies make it clear that Emc protein is synthesized in excess and stabilized in Da heterodimers (Li and Baker, 2018). Accordingly, elevating Da expression also elevates Emc protein levels, primarily increasing levels of inactive Da/Emc heterodimer, potentially without little effect on Da activity. A better route to elevate Da activity is to express tethered Da dimers unable to heterodimerize with Emc (Castanon et al., 2001). Expressing tethered Da dimers using an actin-Gal4 transgene conditionally activated by excision of an FRT cassette (the actin flip-on method) led to tiny clones, reflecting inhibitory effects of Da on growth and survival, and effects on fate specification were difficult to assess. Therefore, we turned to expression under Gal4 control. Expression of Da dimers posterior to the furrow using GMR-Gal4 led to more cone and R7-like cells, but there was also an increase in R8 cell numbers, consistent with a role for Da in R8 specification within the morphogenetic furrow (Fig. 6A-D). Because R8 is directly or indirectly responsible for inducing many ommatidial cell fates, effects on R7 and cone cell numbers could be indirect consequences of excess R8 specification. Accordingly, we expressed da dimers using Lz-Gal4, which is active in the progenitor cell pool that remains after the five-cell R8, R2, R3, R4, R5 preclusters have formed (Crew et al., 1997). This did not affect R7 cell specification but caused disorganization of cone cells in the eye disc and frequent loss of cone cells (Fig. 6E-J) as labelled by the cone cell marker Cut. These eye disc cone cell defects manifest into a rough eye in the adult Drosophila (Fig. 8A,B), similar to eye defects observed by emc knockdown (Fig. 8C). Inter-ommatidial bristles were also abnormal. Simultaneously removing both emc and da using RNAi rescued eye defects to wild type, confirming that defects of emc knockdown were due to Da activity (Fig. 8D).

Having found that emc regulates eye development through Da, and impacts N signaling, we also checked the effects of emc on Sev and Rap1. We report that expression of Sev protein was cell-autonomously reduced within emc mutant clones. Sev is expressed in many cells, not only R7 precursors (Tomlinson et al., 1987), and this overall pattern was affected in emc mutant clones (Fig. 7A,B). Concomitantly, we found that expression of RapGap1 protein, a negative regulator of Rap1 activity, was cell-autonomously elevated within emc mutant clones. RapGap1 protein is expressed posterior to the morphogenetic furrow (Chen et al., 1997). The expression pattern was generally elevated in emc mutant clones (Fig. S7).

To see how altered signaling activities contributed to emc mutant phenotypes, we turned to genetic interactions. Expression of an activated form of Rap1 (Rap1Q63E) (Bonello et al., 2018) rescued the Lz-Gal4 emc knockdown mediated rough eye phenotype, almost to wild type (Fig. 8E,E′). RNAi knockdown of RapGap1 resulted in a similar rescue of Lz-Gal4 emc knockdown (Fig. 8F,F′). Similar suppression was seen for the phenotype of Lz-Gal4 driving Da homodimer expression (Fig. 8G,G′-H,H′). Also, we found that Da overexpression was sufficient to elevate RapGap1 levels (Fig. 9A,B; Fig. S9). In the wild type, RapGap1 protein appeared in the typical R cell differentiation sequence, first in R8, R2 and R5 cells, then in R3 and R4 cells, and finally in the R1, R6 and R7 cell precursors (Fig. 9A). RapGap1 mostly seems to be expressed in R1-8 cells, little or none being detected in other eye disc cells (Fig. S9). Protein appeared mostly cytoplasmic, and prominent in R cell axons and microvilli. These results suggested that reduced Rap1 activity contributed to the emc phenotype. To test this definitively, RapGap1 emc double mutant cells were examined by inducing emc mutant clones in the homozygously viable rapGap1 mutant background. If elevated RapGap1 is contributing to the failure of R7 specification in emc mutants, emc mutant clones should show more normal R7 specification in the rapGap1 mutant background. As predicted, R7 cell specification was rescued substantially in clones of rapGap1 emc double mutant cells (Fig. 9C-H; 30/53 cases). No effect on R8 specification was observed (Fig. 9I). The rapGap1 mutant background alone has no effect on R7 specification (Fig. 9E,G,H).

The emc gene encodes the sole Drosophila representative of the Id protein family, whose members exert diverse effects on development in Drosophila and in mammals. The Id proteins are well known as antagonists of bHLH protein DNA binding and function, including myogenic and proneural genes as well as E proteins (Ling et al., 2014; Roschger and Cabrele, 2017; Wang and Baker, 2015b). Aspects of the emc phenotype in Drosophila affect processes that are independent of these bHLH proteins, raising the possibility of additional mechanisms. How Id proteins can function seemingly independently of its heterodimer binding partners is also a question in mammals, where four Id protein genes exist, and it is difficult to attribute all the phenotypes of individual Id gene knockouts to the many mammalian proneural bHLH genes (Ling et al., 2014; Roschger and Cabrele, 2017; Wang and Baker, 2015a).

We previously showed that the role of emc in proper growth of undifferentiated imaginal disc cells is due to its restraint of the ubiquitous E protein Da, which is thought to homodimerize in the absence of emc (Bhattacharya and Baker, 2011). This is even though imaginal disc growth does not depend on da in the wild type (Bhattacharya and Baker, 2011). A similar conclusion was reached by others (Andrade-Zapata and Baonza, 2014). Here, we report that aspects of emc during eye development also reflect requirements to restrain da, this time in post-mitotic cells, and identify some of the target pathways (Fig. 10).

Among multiple roles of emc during eye development are requirements to prevent R7 precursor cells from adopting R1/6 fate, and for the specification and timely differentiation of cone cells. No proneural bHLH proteins are known to be required for R7 or for cone cell development, and da is also not required for R7 or for cone cell fates. Here we show through analysis of da emc double mutant clones, and also of eyes depleted for both emc and da, that the emc phenotypes depend on da, and must be consequences of da activity when it is not restrained by emc (Fig. 10).

We had not previously observed R7 or cone cell defects after over-expressing da in the eye, which would be predicted to occur. This could be explained by the stabilization of Emc protein by heterodimerization with Da, through which elevated Da expression leads mostly to increased amounts of inactive Emc/Da heterodimer, and little or no increase in Da activity (Li and Baker, 2018). We now report that expression of tethered a Da-Da dimer, which is not subject to inhibition by heterodimerization with Emc, did lead to cone cell defects similar to emc depletion (Figs 7,,8). We do not know whether we have not observed any effect of Da-Da dimers on R7 for technical reasons, or because Da alone is not sufficient for R7 transformation to R1/6, although it is required.

We found previously that emc was required for the proper level and timing of N activity, which is required for both R7 and cone cell fate specification (Bhattacharya and Baker, 2009). Here, we established that N activity in the R7 equivalence group, as measured by E(spl) protein expression, was affected in emc mutants in a da-dependent manner. We also found that two other signaling pathways required for R7 specification, Sevenless and Rap1, depended on emc. Sevenless protein expression levels depended on emc function, and emc was required to restrain levels of RapGap1, a negative regulator of Rap1 activity. The increased RapGap1 levels, expected to diminish Rap1 function in emc mutant cells, were functionally significant, because expression of activated Rap1 suppressed phenotypes of emc depletion or tethered Da-Da expression, and mutation of RapGap1 restored normal R7 specification to emc mutant clones. In addition to the absence of R7 cells from Rap1 mutants, cone cells are disorganized and sometimes missing, resembling the Da dimer expression defects (Mavromatakis and Tomlinson, 2012b) (Fig. 10).

In undifferentiated imaginal disc cells, Emc prevents transcriptional activation of ex, a gene in the Hippo pathway, by Da (Wang and Baker, 2015b). Another role of emc has also since been attributed to unrestrained da activity. The emc gene is required for left–right asymmetrical development of the Drosophila hindgut, in which the potential da target encodes a Myosin, MyoID (Ishibashi et al., 2019). The direct transcriptional target, or targets, of da in the eye are not yet clear. It is interesting that restoring either Rap1 or N activity seems sufficient to restore R7 differentiation. One explanation could be that N, Sev and RapGap1 act in a pathway, for example that N activates RapGap1 expression, which is required in turn for Sev expression. It has been proposed previously that Sev gene expression is likely Notch-dependent, which could potentially account for the reduction in Sev expression in emc mutant clones where N activity is diminished (Tomlinson et al., 2011). Proper Sev protein expression also requires Rap1 activity, possibly through the maintenance of cellular adhesion structures (Baril et al., 2014). RapGap1 expression could be another target of N signaling. However, other hypotheses are also possible. For example, Rap1 has been proposed to maintain adhesion contacts between eye disc cells and therefore to facilitate cell–cell signaling. If RapGap1 is a target of Da, and Rap1 activity required for proper N and Sev signaling, this could also explain why all three pathways are reduced in emc mutant cells, and why restoring Rap1 activity is sufficient to rescue R7 specification. Recently, Rap1 was reported to modulate Notch signaling during eye development (Yost et al., 2023). The preferred E box binding sequence for homodimers of Da and of mammalian E proteins, CACCTG, is short and found at many locations through the genome, so it would not be surprising if unrestrained Da homodimers affected multiple gene targets (Massari and Murre, 2000; Wang and Baker, 2018). A recent study suggests that multiple da-dependent effects of emc mutations are due to elevated expression of the Notch ligand Delta, although it is not known whether this occurs transcriptionally (Nair and Baker, 2023 preprint).

The main conclusion from this study is that restraining the ubiquitous E protein Da can make specification of even cells that do not normally depend on proneural genes dependent on emc. This is in addition to the well-known role of Emc (and Id proteins) regulating processes that do depend on proneural proteins and/or E proteins (Bhattacharya and Baker, 2011; Ling et al., 2014; Troost et al., 2015; Wang and Baker, 2015a). The fate specification examples shown here provide still more examples of an Id protein affecting seemingly bHLH-independent processes by restraining E proteins from novel activity, not through HLH-independent function of the Id protein. This could be a common theme, given the widespread expression of E proteins.

Many molecular features of the Emc/Da system in Drosophila seem to be shared by mammalian Id and E protein genes, including the roles of Id proteins in E protein gene regulation (Richter et al., 2012; Schmitz et al., 2012), and stabilization of Id proteins in heterodimers (Li and Baker, 2018; Wojnarowicz et al., 2019). Notably, the effect of Id2 loss during mammalian NK cell specification is caused by E2a activity, analogous to the role of emc antagonizing da in Drosophila (Boos et al., 2007). Indeed, the mammalian Id1 protein was originally discovered as an inhibitor of E protein function in the mammalian immune system, independent of any proneural-like bHLH proteins, although this is a case where E proteins are required in the wild type. Our results solidify the conclusion that Id proteins have two main modes of action. One is as competitive inhibitors of Ac, Sc, MyoD and perhaps some other bHLH transcription factors with master regulatory roles in specific cell fate decisions. Another is as barriers to transcriptional activation by the E proteins, represented in Drosophila by Daughterless, which are expressed ubiquitously and capable of interfering with many processes when unrestrained.

Mosaic clones were obtained by FLP/FRT-mediated mitotic recombination technique (Golic, 1991; Newsome et al., 2000; Xu and Rubin, 1993). For non-Minute genotypes, larvae were subjected to 1 h heat shock at 37°C at 60±12 h after egg laying and for Minute genotypes, larvae were subjected to heat shock at 84±12 h after egg laying. Larvae were usually dissected ∼72 h after heat shock. Flies were maintained at 25°C.

The following Drosophila strains were used: emc^AP6 (Ellis, 1994); da¹⁰ (Caudy et al., 1988); [UbiGFP] M(3)67C FRT80 (Janody et al., 2004); [Ubi-GFP] FRT40; FRT82 [tub-Gal80] (Lee and Luo, 1999); P{GawB} P{GawB}lz^Gal4 (Crew et al., 1997); UAS- emc ^RNAi (VDRC KK100587) (Dietzl et al., 2007); P{TRiP.JF01766}attP2 (UAS-Ras1Gap RNAi) (Perkins et al., 2015); rapgap1 ²² (Chen et al., 1997); UAS-da-da (Wang and Baker, 2018); and UASpGFP-Rap1Q63E (Ellis et al., 2013). Detailed genotypes are described in the figure legends.

Antibody labeling of eye discs was performed as described previously (Bhattacharya and Baker, 2009; Firth et al., 2006). Images were recorded using BioRad Radiance 2000, Leica SP2 or SP8 Confocal microscope and processed using NIH ImageJ, and Adobe Photoshop software. Primary antibodies used were: mouse anti-Pros 1/25 (MR1A) (Spana and Doe, 1995), rat anti-Elav 1/50 (7E8A10) (Gaul et al., 1992), mouse anti-Cut 1/20 (Schroeder et al., 2011), mouse anti-Da (1:200) (Cronmiller and Cummings, 1993), guinea pig anti-Runt 1/1500 (Duffy et al., 1991), guinea pig anti-Sens 1/500 (Nolo et al., 2000), anti-E(spl) (mAb323) 1/1 (Jennings et al., 1994), mouse anti-Svp 1/1000 (Kanai et al., 2005), anti- Sevenless (Tomlinson et al., 1987), anti-RapGap1 (m4G5H3) 1/4 (Chen et al., 1997), and anti-GFP 1/500 (Invitrogen), Secondary antibodies used were multilabeling antibodies from Jackson ImmunoResearch Laboratories. To quantify RapGap1 levels, pixel intensity from maximum projections was measured in Fiji. Regions posterior to the SMW (estimated from the ELAV pattern) were compared after normalizing to control discs labelled in parallel.

Sample sizes were determined based on prior experience in similar experiments. Experiments were not blinded. No data were excluded unless parallel wild-type controls indicated technical failure.

We thank J. Benitez for experimental help, P. M. Domingos, P. Heitzler, R. Mann, M. Peifer, R. Kellum, I. Hariharan, L-H. Wang and the National Drosophila Stock Center at Bloomington for Drosophila strains, C. Cronmiller, P. M. Domingos, Y. N. Jan, H. Bellen, S. Bray, Marc Therrien, I. Hariharan, C. Desplan and the Developmental Studies Hybridoma Bank for antibodies. We thank J. Secombe and L-H. Wang for comments on the manuscript. Data in this paper are from a thesis submitted in partial fulfilment of the requirement for the degree of Doctor of Philosophy in the Graduate Division of Biomedical Sciences, Albert Einstein College of Medicine, Yeshiva University, USA. Confocal microscopy and Electron microscopy (SEM) was performed at the AIF, AECOM. The Leica SP8 microscope was acquired through NIH SIG 1S10 OD023591. We thank L. Gunther-Cummins and X. Nishku for SEM help.