Inhibition of Daughterless by Extramacrochaetae mediates Notch-induced cell proliferation

An essential component of development is the regulation of cell proliferation. All tissues and organs originate from just a handful of early embryonic cells. These cells must divide rapidly in order to produce the requisite number of cells for the assigned tissue fate. As organs vary considerably in their final size, each precursor cell population must adjust its growth rate in order to develop synchronously with other tissues/organs. In addition, each organ must have a way to sense when its final size has been reached and then employ mechanisms to inhibit further cell proliferation. These processes are regulated by a multitude of signal transduction cascades and transcription factor networks, many of which directly impinge on the cell cycle and/or the apoptotic machinery.

Growth of the developing eye is controlled by two different phases of cell proliferation. The first phase can be broadly defined to include all cell divisions that take place during the first and second larval instars, as well as those ahead of the morphogenetic furrow during the third and final larval stage. This includes a coordinated wave of mitosis that occurs just ahead of the morphogenetic furrow called the ‘first mitotic wave’. This early growth phase is responsible for setting the overall number of ommatidia within the compound eye and is controlled by several signaling pathways, including JAK/STAT, Notch, Decapentaplegic (Dpp/TGFβ), Wingless (Wg) and Hippo (Panin et al., 1997; Cho and Choi, 1998; de Celis et al., 1998; Domínguez and de Celis, 1998; Go et al., 1998; Papayannopoulos et al., 1998; Kango-Singh et al., 2002; Tapon et al., 2002; Baonza and Freeman, 2005; Firth and Baker, 2005; Reynolds-Kenneally and Mlodzik, 2005; Fan and Bergmann, 2008; Gutierrez-Aviño et al., 2009; Koontz et al., 2013). The overlying peripodial membrane serves as an additional source of growth signals. Several studies have demonstrated that Dpp, Wg, Hedgehog (Hh) and Serrate (Ser), a Notch ligand, are all transported from the peripodial membrane and are required for the growth (and patterning) of the eye disc (Cho et al., 2000; Gibson and Schubiger, 2000; Gibson et al., 2002).

The second growth phase comprises the ‘second mitotic wave’, a synchronized band of mitoses that occurs behind the furrow and produce the following elements of each ommatidium: three of the eight photoreceptors (R1,6,7), all lens-secreting cone cells, all optical insulating pigment cells and all cells of the bristle complex (Ready et al., 1976; Wolff and Ready, 1991). These late cell divisions produce the vast majority of cells within each ommatidium and are therefore the major determinants of the overall number of cells within the adult retina. The Notch and EGF Receptor (Egfr) pathways regulate entry into and passage through the cell cycle of the second mitotic wave (Baker and Yu, 2001; Baonza et al., 2002; Yang and Baker, 2003; Baonza and Freeman, 2005; Firth and Baker, 2005, 2007; Sukhanova and Du, 2008). In addition to these growth signals, the Hh pathway controls compensatory proliferation as cells differentiate behind the furrow (Fan and Bergmann, 2008).

The establishment of the midline within the developing eye is essential for Notch activation and cell proliferation during early growth phases (Cho and Choi, 1998; Domínguez and de Celis, 1998; Papayannopoulos et al., 1998). The embryonic and first instar eye is composed of entirely ventral-fated tissue (Chern and Choi, 2002; Singh and Choi, 2003; Singh et al., 2006). These cells express fringe (fng), which encodes a glycosyltransferase (Irvine and Wieschaus, 1994). In the latter half of the first instar, pannier (pnr) expression is activated along the dorsal margin of the eye and in turn Pnr protein initiates a cascade of events that leads to the repression of fng with the dorsal half of the retina (Heitzler et al., 1996; Cho and Choi, 1998; Heberlein et al., 1998; Cavodeassi et al., 1999; Yang et al., 1999; Maurel-Zaffran and Treisman, 2000; Sato and Tomlinson, 2007; Oros et al., 2010). The juxtaposition of fng-positive and fng-negative cells leads to the activation of Notch signaling at the midline, which plays an essential role in promoting localized growth in the eye (Panin et al., 1997; Cho and Choi, 1998; Domínguez and de Celis, 1998; Papayannopoulos et al., 1998).

Here, we address the role that the helix-loop-helix (HLH) protein Extramacrochaetae (Emc) plays in mediating some of the Notch-induced growth. The emc locus was first identified by mutations that increased the number of macrochaetae class bristles (Craymer, 1980). emc, along with hairy (h), was subsequently shown to genetically modify the mutant phenotypes of members of the achaetae-scute complex (AS-C) (Botas et al., 1982; Moscoso del Prado and Garcia-Bellido, 1984). The diametrically opposing phenotypes of emc and AS-C loss-of-function mutants hinted at a biochemically antagonistic relationship. Members of the AS-C encode basic helix-loop-helix (bHLH) transcription factors, whereas Emc encodes a helix-loop-helix (HLH) protein (Villares and Cabrera, 1987; Murre et al., 1989a,b; Ellis et al., 1990; Garrell and Modolell, 1990). Emc binds to and forms heterodimers with several AS-C proteins and the class I bHLH factor Daughterless (Da) (Van Doren et al., 1991, 1992; Alifragis et al., 1997). However, as Emc lacks a basic 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 downstream target genes. Mutations within emc have documented growth defects in both the developing wing and eye (García-Alonso and García-Bellido, 1988; de Celis et al., 1995; Baonza and García-Bellido, 1999; Baonza et al., 2000; Bhattacharya and Baker, 2009; Spratford and Kumar, 2013). However, the sequestration targets of Emc and the developmental mechanism by which it promotes cell proliferation are not well defined.

In several contexts, emc appears to function downstream of the Notch pathway (Baonza et al., 2000; Baonza and Freeman, 2001; Adam and Montell, 2004; Tapanes-Castillo and Baylies, 2004; Bhattacharya and Baker, 2009; Spratford and Kumar, 2015). Here, we show that the Notch pathway activates emc expression. Emc protein then promotes cell proliferation by binding to and sequestering Da away from endogenous DNA targets. We show that the sequestration of Da by Emc is sufficient to block the growth-inhibiting activity of Da. And finally, we provide evidence that entry into S phase may be slowed in emc mutant tissue. Our study complements another study that indicates that Emc-mediated regulation of Da also affects passage of dividing cells through the G2/M checkpoint (Andrade-Zapata and Baonza, 2014).

Several studies have demonstrated that emc mutant clones proliferate poorly in comparison with wild-type cells (García-Alonso and García-Bellido, 1988; de Celis et al., 1995; Baonza and García-Bellido, 1999; Baonza et al., 2000; Bhattacharya and Baker, 2009). In this paper, we set out to determine the mechanism by which Emc regulates tissue growth in the developing eye disc. We began this study by measuring the growth rates of clones of wild-type and emc^AP6 null mutant cells. Consistent with prior reports, we observe that the growth rate of emc^AP6-null mutant clones is markedly less robust when compared with wild-type clones (Fig. 1A,B,E,F,I,J,M). However, if emc^AP6-null clones are surrounded by tissue that is heterozygous for a Minute mutation, then the mutant tissue grows just as well as wild-type cells that are placed in the same environment (Fig. 1C,D,G,H,K,L,M). From these data we conclude that the defects in tissue growth are not due to a requirement for Emc in maintaining cell viability. We next considered a role for Emc in the suppression of apoptosis. To test this hypothesis, we used the MARCM method to overexpress p35 in both wild-type and emc^AP6 mutant cells (Fig. 1N-R). We did not observe an increase in cleaved caspase 3 (CC3) levels in emc^AP6 mutant clones (data not shown) nor did the expression of p35 restore growth to emc^AP6 mutant cells (Fig. 1R); therefore, we conclude that the growth deficit in emc mutant tissue is also not due to an increase in apoptotic cell death. In light of this set of conclusions, we turned our attention to a possible role for Emc in promoting cell proliferation.

We focused on a potential connection between Notch signaling and Emc as several studies have demonstrated that Emc functions downstream of the Notch pathway in the wing and eye discs, embryonic mesodermal segments, and ovarian follicle cells (Baonza et al., 2000; Adam and Montell, 2004; Tapanes-Castillo and Baylies, 2004; Bhattacharya and Baker, 2009; Spratford and Kumar, 2015). In fact, emc expression appears to be dependent upon Notch pathway activity in several different contexts (Baonza et al., 2000; Baonza and Freeman, 2001; Bhattacharya and Baker, 2009; Spratford and Kumar, 2015). Activation of an emc transcriptional reporter (emc-lacZ) by overexpression of an activated Notch receptor (N^ICD) in the eye (Fig. 2A-D) is consistent with these findings. In an attempt to identify regulatory regions that may be responsive to Notch signaling, we analyzed the expression of eight genomic fragments surrounding the emc transcriptional start site (supplementary material Fig. S1A; Jory et al., 2012; Manning et al., 2012). Expression within the anterior compartment was of particular interest, as a significant level of Notch-dependent cell proliferation takes place within this region. Of the eight fragments that were tested, we found two putative enhancer regions, including E1 and GMR10H11, that have anterior expression (supplementary material Fig. S1B,F). Within both fragments are binding sites for Suppressor of Hairless [Su(H)], a transcription factor responsible for mediating Notch signaling (data not shown).

We tested whether emc is required for the proliferative effects of Notch signaling in the eye disc by creating and comparing MARCM clones that overexpress N^ICD in wild-type and emc^AP6 mutant clones. We find that the proliferative effects of Notch are partially mitigated by the loss of emc (Fig. 2E-G). From these results, we conclude that Emc mediates some of the Notch-induced proliferation in the eye disc. We did observe that the loss of emc was not sufficient to completely abrogate the effects of Notch signaling on growth (Fig. 2G). One plausible explanation for this result is that Notch and Emc may act on some common target genes. A similar phenomenon has been observed in the wing (Baonza et al., 2000; Campuzano, 2001).

As Emc lacks a basic domain, neither Emc itself nor Emc-bHLH heterodimers should be capable of interacting with DNA (Ellis et al., 1990; Garrell and Modolell, 1990; Van Doren et al., 1991). Thus, Emc and its vertebrate homologs are thought to influence transcription by sequestering bHLH proteins away from target genes. In particular, genetic interaction studies and electromobility shift assays (EMSAs) have identified Da and AS-C members as targets of Emc sequestration (Van Doren et al., 1991, 1992; Martinez et al., 1993). Based on this mode of action, Emc is predicted to promote cell proliferation by sequestering one or more bHLH proteins that, on their own, are tasked with repressing tissue growth. One of the unresolved issues surrounding Emc is the identification of its sequestration targets in proliferating cells. We initiated a yeast two-hybrid screen for Emc-binding partners and identified several candidate bHLH proteins, including Da, Ac and Sage, a bHLH protein that is expressed exclusively in the salivary glands (supplementary material Fig. S2). We focused our attention on Da because flp-out overexpression clones display a severe undergrowth phenotype (Bhattacharya and Baker, 2011; Andrade-Zapata and Baonza, 2014), which is the expected phenotype if the sequestration of Da by Emc is required to promote tissue growth. We conducted a directed yeast two-hybrid assay to assess the strength of the interaction between Emc and Da, and find that it is equivalent to our medium/strong positive control colonies (Fig. 3A). We then confirmed that Emc could sequester Da away from a target sequence. To do this, we developed a luciferase transcriptional reporter assay that could be used in transfected Kc167 Drosophila cells. We multimerized an E-box sequence (Massari and Murre, 2000; Materials and Methods) and placed this synthetic regulatory element within a vector upstream of the firefly luciferase gene. This plasmid was transfected into Kc167 cells simultaneously with combinations of the following plasmids: (1) MT-GAL4, (2) UAS-Da, (3) UAS-Emc, (4) UAS-GFP and (5) UAS-renilla luciferase. Da, on its own, is capable of binding to and strongly activating the E-box-luciferase reporter (Fig. 3B). The addition of increasing amounts of Emc significantly inhibits reporter activation, thereby supporting the proposition that Emc antagonizes Da (and presumably other bHLH proteins) through sequestration (Fig. 3B).

We also tested whether Emc could interfere with Da activity through additional mechanisms. In particular, we focused our efforts on determining whether the Emc-Da heterodimer could still bind to DNA as an inactive complex and/or whether Emc could directly interact with DNA, thus preventing bHLH proteins from binding to the same site. To test the first model, we fused the strong activation domain of VP16 to Emc (Emc-VP16) and compared the ability of Emc-VP16 to modulate the transcriptional activity of Da. If Emc and Da are capable of interacting with DNA as a heterodimer, then the Da-Emc-VP16 complex should activate the reporter at higher levels than Da alone. Our results show a significantly decreased level of activation instead, which indicates that the Da-Emc complex does not interact with DNA as a heterodimer (Fig. 3C). To test the ability of Emc to directly compete for DNA-binding sites with Da we compared the ability of wild type Emc, Emc-VP16 and GFP-VP16 (a negative control) to activate the E-box luciferase reporter. Wild-type Emc failed to activate the reporter above basal levels and we observed no difference between Emc-VP16 and GFP-VP16 (Fig. 3D). Last, an Emc-GAL4^DBD chimeric protein is unable to activate transcription of several reporters that were placed under UAS control in yeast cells (data not shown). Together, these results suggest that Emc itself completely lacks transcriptional activation potential and does not compete with Da (or other bHLH proteins) for E-box sites in the genome. The sequestration of bHLH proteins appears to be the only method by which Emc antagonizes Da.

To test whether the sequestration of Da by Emc is required to promote growth in the developing eye, we generated and measured the size of clones that overexpress emc, da or both genes together. Although our heat-shock regime (see Materials and methods), is sufficient to induce healthy wild-type clones (Fig. 4A,E), we failed to recover any da overexpressing clones (Fig. 4B,E). This suggests that Da normally inhibits tissue growth and it is consistent with the undergrowth phenotype that has been previously observed with its overexpression (Bhattacharya and Baker, 2011). Growth is restored to da overexpression clones by the co-expression of emc (Fig. 4D,E), indicating that Emc can sequester Da in vivo to promote cell proliferation. Last, overexpression of emc, on its own, is also sufficient to induce clones that are significantly larger than da-expressing clones (Fig. 4C,E). In this case, it is likely that the exogenous Emc is promoting growth by sequestering endogenous Da. Finally, we attempted to generate MARCM clones in which da is overexpressed within emc^AP6-null mutant clones but these clones fail to survive. This is consistent with the undergrowth phenotype that is observed when da is overexpressed by itself. Taken together, our data suggest that the sequestration of Da by Emc is required to promote cell proliferation in the eye.

It has been reported that the removal of Notch signaling (using a temperature-sensitive allele) abolishes the high level of da expression that exists between neighboring proneural clusters (Lim et al., 2008). Consistent with this result, the hyper-activation of Notch signaling in MARCM clones leads to an increase in da expression (supplementary material Fig. S3A-C). Notch activation of da appears to be dependent upon emc, as the activation of Notch signaling in emc^AP6-null clones failed to activate da expression (supplementary material Fig. S3D-F). The apparent requirement for Emc in Notch-mediated activation of da is at odds with another report in which Emc was shown to repress da expression (Bhattacharya and Baker, 2011). The authors demonstrated that, in emc mutant clones, da expression was elevated compared with surrounding wild-type tissue. Although the reasons underlying these conflicting results are not clear, our results are consistent with a model in which Notch-mediated growth requires Emc to sequester Da.

In addition to Da, three other bHLH proteins (Myc, Max and Mnt) are also important for regulating cell proliferation and differentiation (Grandori et al., 2000; Orian et al., 2003) and thus could be potential sequestration targets of Emc. Similar to da, clones that overexpress Mnt are rarely recovered and are small in size (Loo et al., 2005). We overexpressed Mnt throughout the entire eye using an ey-GAL4 driver and observe a range of rough eye phenotypes (Fig. 5A-D, 136 females and 117 males were analyzed). The small rough eye (Fig. 5C) is consistent with earlier efforts to overexpress Mnt in the eye (Loo et al., 2005). We were able to partially suppress the eye defects by co-expressing emc (Fig. 5D, 142 females and 112 males were analyzed). Expression of emc alone does not alter the structure or size of the compound eye (Fig. 5D, 156 females and 147 males analyzed) and is indistinguishable from controls (Fig. 5D, 180 females and 202 males analyzed). Based on this genetic interaction, we used directed yeast two-hybrid assays and immunoprecipitation in rabbit reticulocyte lysates to test for physical interactions (and thus potential sequestration) between the two proteins but we failed to detect Emc-Mnt complexes in either assay (Fig. 5E, data not shown).

Myc, which is encoded by the diminutive (dm) gene, is an important player in cell proliferation and cell competition in several Drosophila tissues, including the wing and eye (Johnston et al., 1999; de la Cova et al., 2004). The overexpression of Myc throughout the eye, unlike Mnt, does not alter the overall size and structure of the adult retina (Bellosta et al., 2005); therefore, a simple genetic interaction assay could not be performed. However, we did examine emc^AP6-null mutant clones and observed that the level of Myc protein is decreased (Fig. 5F-H). The reduction in Myc levels suggests that the undergrowth phenotype of emc mutant tissue may be due in part to reductions in Myc levels. A complete knockdown of Myc is seen only in a subset of emc mutant clones. The partial knockdown seen in other clones may suggest that additional inputs (independent of Emc) regulate Myc expression. A transcriptional link between Myc and Emc/Id also appears to exist in vertebrates, but interestingly in this case Myc appears to lie upstream and activates the expression of Id2, a homolog of Emc (Lasorella et al., 2000). Despite these differences, the data from both systems suggest that there is a transcriptional chain linking Myc and Emc/Id. To test whether Emc also binds and sequesters Myc, we used directed yeast two-hybrid and immunoprecipitation assays to probe for the formation of an Emc-Myc protein complex. We failed to detect such a complex (Fig. 5E, data not shown); thus, it is likely that Emc and Myc are connected mainly via transcriptional regulation and not through protein sequestration.

In mammalian systems, Id2 binds to Retinoblastoma (Rb) protein (Lasorella et al., 2000). This is consistent with Emc being capable of binding to non-bHLH proteins in Drosophila (Giot et al., 2003). Sequestration of vertebrate Rb or its two fly homologs, Rbf1 and Rbf2, (Du et al., 1996; Stevaux et al., 2002) is predicted to increase the levels of E2F that would be available to promote proliferation through the transition into S phase (Nevins, 1992a,b; Weinberg, 1995; Sherr, 1996). Both Rbf1 and Rbf2 were tested for interactions with Emc via a directed yeast two-hybrid assay and co-immunoprecipitation in rabbit reticulocyte lysates but were not found to interact (supplementary material Fig. S4). We also used a PCNA::GFP reporter, which contains several E2F-binding sites as readout for E2F levels. The PCNA::GFP reporter remains unchanged in clones that overexpress emc (Fig. 6Q-T). If Emc bound to either Rbf1 or Rbf2 in vivo then we would expect to see increased levels of PCNA::GFP as increased levels of E2F would be available to bind and activate the reporter. However, the data from our biochemical studies and the PCNA::GFP reporter support the overall conclusion that Emc does not regulate Rb/E2F, but instead promotes proliferation by sequestering/inhibiting Da and by genetically interacting with Myc and Mnt via mechanisms that are unlikely to involve their direct protein sequestration.

As emc mutant tissue grows poorly, we set out to determine whether entry into or passage through the cell cycle is negatively impacted. To do this, we analyzed the expression profiles of several cell cycle markers in emc^AP6 loss-of-function clones. Most of the cell cycle regulators that we tested appear normal (Fig. 6A-L). However, we did notice that EdU incorporation is reduced in emc^AP6 clones (Fig. 6M-P; supplementary material Fig. S5A-F). An analysis of 48 clones anterior to the furrow indicates that in 22.9% of emc^AP6 clones (11/48) there is an observable reduction in EdU incorporation when compared with surrounding wild-type tissue. EdU incorporation in the remaining emc^AP6 clones is comparable with wild-type cells. These results suggests that entry into S phase may be either delayed or partially blocked, and that, in normal cells, Emc may participate in the entry into or influence the length of time that is spent in S phase. Such a role for Emc at this point in the cell cycle would be consistent with the known role that Notch signaling plays in cell cycle progression through the second mitotic wave (Baonza and Freeman, 2005; Firth and Baker, 2005; Sukhanova and Du, 2008). A delay or partial block in the entry into S phase is also consistent with the undergrowth phenotype that is associated with emc loss-of-function mutant tissue and da overexpression clones.

Cells lacking emc have been reported to grow poorly in several developing tissues, including the ovary, wing and eye (de Celis et al., 1995; Baonza and García-Bellido, 1999; Baonza et al., 2000; Adam and Montell, 2004; Bhattacharya and Baker, 2009; Spratford and Kumar, 2013; Andrade-Zapata and Baonza, 2014). In this manuscript, we have attempted to bring several observations together and demonstrate that Emc mediates some of the Notch-mediated growth in the developing eye by sequestering the type I bHLH protein Da. Several previous findings are crucial for this study: first, in several developmental contexts, emc appears to lie downstream of the Notch signaling pathway (de Celis et al., 1995; Baonza and García-Bellido, 1999; Baonza et al., 2000; Adam and Montell, 2004; Bhattacharya and Baker, 2009; Spratford and Kumar, 2015). This is an important observation as Notch signaling is a key promoter of growth in the eye (Cho and Choi, 1998; de Celis et al., 1998; Domínguez and de Celis, 1998; Go et al., 1998; Papayannopoulos et al., 1998; Kenyon et al., 2003; Chao et al., 2004; Dominguez et al., 2004; Baonza and Freeman, 2005; Firth and Baker, 2005; Reynolds-Kenneally and Mlodzik, 2005). Second, Emc antagonizes the activity of several bHLH proteins, including Da, by binding to and sequestering them away from target DNA sequences (Van Doren et al., 1991, 1992; Martinez et al., 1993; Alifragis et al., 1997). Finally, the overexpression of Da, leads to an undergrowth phenotype similar to that seen with emc loss-of-function mutants (Bhattacharya and Baker, 2011).

We started this study by analyzing the proliferative state of cells that lack emc. Although cells lacking emc grow very poorly, the mutant cells are not subviable. When placed in competition with cells that have been given a growth disadvantage, emc mutant cells grow as well as wild-type cells (Fig. 1A-M). This undergrowth phenotype does not appear to be due to an increase in apoptotic levels, as the expression of p35, which blocks all caspase-dependent cell death, cannot restore normal growth rates to emc mutant cells (Fig. 1N-R). Thus, emc mutant cells are neither intrinsically compromised in their ability to survive nor do they undergo apoptotic cell death. Instead, it appears that cells lacking emc are deficient in the ability to proliferate. As emc has been implicated in Notch-mediated growth in the developing wing, we sought to determine whether a similar relationship exists within the eye. To accomplish this goal, we analyzed the growth of three different types of cell populations: (1) emc-null mutant cells, (2) wild-type cells overexpressing activated Notch and (3) emc-null mutant cells overexpressing activated Notch. The proliferative effects of Notch are greatly diminished in emc mutant cells (Fig. 2E-G) – this indicates that, as in the wing, Emc mediates some of the Notch-dependent cell proliferation in the eye. Interestingly, the loss of emc does not completely abrogate Notch-mediated growth (Fig. 2E-G), suggesting that the Notch pathway may have targets that are not regulated by Emc.

We set out to determine the mechanism by which Emc modulates Notch-dependent growth in the eye. Current models suggest that several growth-promoting genes such as four-jointed, cyclin A, cyclin B and E2f1 are directly activated by the Notch pathway via Su(H)/Mam and/or the E(spl) complex (Knoblich and Lehner, 1993; Baonza and Freeman, 2005; Gutierrez-Aviño et al., 2009). Our data hint at an additional mechanism in which the Notch pathway activates emc expression and, in turn, Emc protein promotes cell proliferation by binding to and sequestering one or more growth-inhibiting bHLH proteins. We have demonstrated here that Notch signaling can activate emc expression in proliferating cells ahead of the morphogenetic furrow (Fig. 2A-C; Spratford and Kumar, 2015). These results are similar to those obtained by others in the wing and in developing photoreceptor neurons. We did not test the possibility that Su(H)/Mam directly binds to the emc locus and activates its expression, but it is an attractive model as two Su(H)-binding sites exist close to the transcriptional start site and emc expression is lost in Su(H)- and mam-null mutant clones (Spratford and Kumar, 2015).

Once emc expression is activated by Notch signaling, we propose that Emc protein goes on to promote growth by sequestering Da. Clones that overexpress da grow very poorly, suggesting that Da functions to inhibit cell proliferation (Bhattacharya and Baker, 2011; Fig. 4A,B,E). This is the expected growth phenotype for a target of Emc and therefore makes Da an ideal candidate for sequestration by Emc in proliferating cells. Consistent with this model, Emc can bind to Da in yeast two-hybrid assays and can interfere with the ability of Da to interact with DNA in electromobility shift assays and in transcriptional activation assays conducted in Drosophila Kc167 cells (Van Doren et al., 1991, 1992; Alifragis et al., 1997; Fig. 3; supplementary material Fig. S2). Here, we further demonstrate that the Emc-Da interaction is an important step in tissue growth, as the expression of emc within the eye disc is sufficient to block the growth inhibiting properties of da overexpression (Fig. 4A-E). Consistent with earlier in vitro protein-DNA binding assays, our transcriptional activation assays in yeast and Drosophila Kc167 cells indicate that protein sequestration appears to be the major, if not only, mechanism by which Emc antagonizes Da (Fig. 3).

Last, we attempted to determine how Emc regulates the cell cycle in the eye disc. Multiple studies have demonstrated that Notch signaling and Da exert opposing activities on the G1/S transition. In particular, Notch signaling promotes entry into S phase, while Da counteracts this activity by activating dacapo (dap) (Fig. 7; Baonza and Freeman, 2005; Firth and Baker, 2005; Sukhanova et al., 2007; Sukhanova and Du, 2008). In addition, da inhibits the transcription of string (stg) (Andrade-Zapata and Baonza, 2014). String, the Drosophila homolog of yeast cdc25, plays an important role in transitioning cells through the G2/M checkpoint in Drosophila (Edgar and O'Farrell, 1989). These observations, coupled with our genetic studies placing Emc between the Notch pathway and Da, suggest that both the G1/S and G2/M transitions may be the points in the cell cycle that are controlled by Emc. Consistent with this model, we find that EdU incorporation is lower in emc mutant cells when compared with wild-type cells (Fig. 6M-P; supplementary material Fig. S5). As emc mutant cells can proliferate when surrounded by growth-deficient cells, it is unlikely that these cells are completely blocked from entering S phase. Rather, it is more likely that emc mutant cells are delayed in S-phase entry. In further support of a role for Emc in regulating G1/S, we find that levels of Myc, which is required for entry and progression through S phase, are severely reduced in emc mutant clones (Fig. 5F-H). A role for Emc in regulating the G2/M transition is proposed in a recent study in which the authors found that emc mutant cells remain in G2 (Andrade-Zapata and Baonza, 2014). In our study, we analyzed cell cycle profile of emc-null mutant cells using a variety of markers. In the other study, the authors used RNAi lines targeting emc and da overexpression constructs to manipulate protein levels for FACS analysis. It is likely that the use of different methods in the two studies have allowed for the identification of roles for Emc at the two different cell cycle checkpoints.

Vertebrate genomes harbor four homologs of emc and are referred to as inhibitors of differentiation 1-4 (Id1-Id4). The expression of the Id genes is mediated by several oncoproteins and is upregulated in numerous cancers (reviewed by Perk et al., 2005). Multiple Id proteins themselves are also known to regulate cell proliferation and tissue growth (Norton, 2000; Zebedee and Hara, 2001; Sikder et al., 2003; Wong et al., 2004; Iavarone and Lasorella, 2006; Ling et al., 2006). We have demonstrated that Emc promotes cell proliferation via the sequestration of Da. Interestingly, all four Id proteins form biochemical complexes with E12 and E47, the vertebrate homologs of Da (Cochrane et al., 2009; Sun et al., 2009; Teachenor et al., 2012). A significant advance will be made if it emerges that the sequestration of Da/E2A by Emc/Id is a universal mechanism by which cell proliferation and tissue growth is promoted in both flies and mammals. It will also be interesting to determine the extent to which sequestration of bHLH proteins by Emc/Id is used during development. In addition to the known sequestration targets, our Y2H screen also identified Sage, a bHLH protein that is specifically expressed in the salivary gland. A comprehensive analysis of potential bHLH-Emc/Id interactions could shed considerable light on the extent to which Emc/Id regulates bHLH-dependent processes in both flies and vertebrate systems.

Fly stocks used were: (1) hs-FLP²²; (2) y¹ w* hs-FLP²²;; FRT80B emc^AP6/TM6B; (3) y¹ w* hs-FLP²²;; FRT80B Ubi-GFP M(3)ⁱ⁵⁵/TM6B; (4) w¹¹¹⁸; FRT80B Ubi-GFP(S65T)nls/TM3 Sb¹; (5) w*; FRT80B ry⁵⁰⁶; (6) y¹ w* hs-FLP²² UAS-mCD8::GFP.L Ptp4E^LL4; Pin¹/CyO; (7) y¹ w*; Act5C-GAL4/CyO; (8) y¹ w*; FRT80B tub-GAL80; (9) UAS-p35; (10) UAS-N^ICD; FRT80B tub-GAL80; (11) y¹ w*; AyGAL4 UAS-GFP(S65T) Myo31DF; (12) w¹¹¹⁸; (13) UAS-da; (14) UAS-emc4M; (15) UAS-da;UAS-emc4M; (16) ey-GAL4; (17) UAS-Mnt^T233; (18) UAS-emc; UAS-Mnt^T233; (19) PCNA. ΔNhe::GFP; and (20) w¹¹¹⁸; Act5C-GAL4 UAS-RFP.W/TM3 Sb¹. A list of full genotypes to accompany the figures can be found in supplementary material Table S1.

Primary antibodies used were: (1) rat anti-Elav (1:100, DSHB); (2) mouse anti-Myc (1:100, DSHB); (3) mouse anti-cyclin A (1:4, DSHB); (4) mouse anti-cyclin B (1:4, DSHB); (5) mouse anti-cyclin E (1:4, DSHB); (6) rabbit anti-pH3 (1:20,000, Abcam, ab32107); and (7) mouse anti-β-gal (1:250, Promega, Z3781). Fluorophore-conjugated secondary antibodies and phalloidin-fluorophore conjugates were obtained from Jackson Laboratories and Molecular Probes. Imaginal discs were prepared as described previously (Anderson et al., 2012). All eye-antennal discs were photographed on a Zeiss Axioplan II compound microscope at 10× magnification, except for the discs in supplementary material Fig. S5, which were photographed at 40×. For scanning electron microscopy, adult flies 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 SEM. All adult heads were photographed at 180× magnification.

This assay was performed using the Life Technologies Click-iT EdU Alexa Flour 594 Imaging Kit. Larval heads were dissected in phosphate buffer. Heads were individually incubated in 50 µM solution of EdU in phosphate buffer for 15 min. Heads were washed with phosphate buffer for 15 min then fixed individually with a 4% paraformaldehyde solution for 30 min. Fixed heads underwent three 5 min washes in 0.3% Triton, one 20 min wash in 0.6% Triton, followed by two 5 min washes in 3% BSA. The Click-iT reaction step was performed for 30 min followed by one 5 min wash in 3% BSA, then two 5 min washes in 0.1% Triton. Eye imaginal discs were then dissected from the head complex and treated as described in the antibodies and microscopy section

Adobe Photoshop CS6 was used to outline and measure the area of the eye imaginal disc (in pixels). Next, within this area, a color range selection was made to select all clonal tissue (GFP positive for flp-out and MARCM clones, and GFP negative for loss-of-function clones). The area of the clones was also measured in pixels. The clonal area was divided by the total area of the eye disc and then multiplied by 100 in order to calculate the percentage of the eye field that is occupied by clonal tissue. The individual percentages for each genotype were then averaged. Statistical significance was calculated using Student's t-test and equal or unequal variance was determined using a F-test. Error bars within the figures represent s.d.

All heat-shock clones were induced at 37°C 48 h after egg lay (AEL). emc^AP6 loss-of-function clones in wild-type or Minute⁻ backgrounds were induced by a 20 min heat pulse followed by a 60 min rest period (room temperature) and a second 20 min heat pulse. Clones in Fig. 1E-L were induced with a single 20 min heat pulse. Overexpression and MARCM clones were induced with 20 and 60 min heat pulses, respectively.

The three putative enhancer regions shown in supplementary material Fig. S1A were amplified from w¹¹¹⁸ genomic DNA using PCR and cloned into the placZ.attB plasmid (a gift from Konrad Basler, University of Zurich, Switzerland). PCR conditions and primer sequences, cloning strategies and sequence files for each construct are all available upon request. The p.enhancer.lacZ.attB constructs were transformed into flies using the phiC31 integrase system. Stable stocks were created and analyzed for lacZ expression in imaginal and brain tissue.

The ProQuest Yeast 2-Hybrid System (Invitrogen) was used to identify proteins that physically associate with Emc. A library of cDNAs from Drosophila third instar larvae was cloned into the pDEST-22 vector, which contains the GAL4 activation domain (prey plasmid). Yeast MaV203 cells were transformed with the prey library and the bait plasmid pDEST-32 in which Emc is fused to the GAL4 DNA-binding domain. The IU Yeast Two-Hybrid Facility conducted this unbiased Y2H screen. Interactions between prey proteins and Emc were identified by the activation of three reporters (UAS-lacZ, UAS-HIS3 and UAS-URA3). Plasmids from positive clones were isolated and sequenced.

Da, Myc, Mnt, Max, Rbf1 and Rbf2 cDNAs were individually cloned into the pDEST22 vector from Invitrogen to create corresponding p22 prey vectors. These were individually tested in pairwise transformations with the p32-Emc bait vector. The strong, weak and negative controls used are listed within the Invitrogen manual. Cultures were grown at 30°C in 500 ml SC-L-T+g media shaking at 200 rpm. After 48 h, 100 ml of each culture was diluted in 2 ml sterile water and replica stamped onto the following plate types: (1) SC-L-T+g, (2) SC-L-T-H+g, (3) SC-L-T-U+g, (4) SC-L-T-H+g 10 mM 3AT and (5) SC-L-T-H+g 100 mM 3AT. Plates were incubated at 30°C. After 1 week, colony growth was recorded and photographed. A da cDNA was also cloned into the p32 bait vector and tested against the Myc-, Mnt- and Max-containing p22 prey vectors.

An oligomer consisting of five repeats of the E-box CANNTG with spacers of ATATAT was constructed by Integrated DNA Technologies (IDT) and cloned into a plasmid containing the hsp70 promoter and firefly luciferase gene. da and emc cDNAs were cloned into vectors containing UAS regulatory sites. The Qiagen Effectene Transfection Reagent was used to transfect K_C167 cells with the E-box reporter plasmid, the pMT-GAL4 driver plasmid, the UAS-Renilla control plasmid and differing combinations of UAS-GFP, UAS-da or UAS-emc plasmids. Transfected cells were incubated at 37°C for 24 h, at which time protein production was induced by adding CuSO₄. Twenty-four hours later, cells were treated according to the Promega Dual-Luciferase Reporter Assay System protocol, and firefly luciferase and renilla luciferase levels were measured using a Glomax 20/20 1996 Luminometer. For each experiment, three independent biological samples were taken. In each case, the firefly luciferase reading was divided by the renilla luciferase control reading to obtain a ratio. The ratios from the three samples were then averaged. Statistical significance was calculated using Student's t-test and equal or unequal variance was determined using a F-test. Error bars within the figures represent s.d.

We thank Brandon and Bonnie Weasner for technical support with molecular cloning and biochemical assays, Bonnie Weasner and Laurel Bender for technical help with directed yeast two-hybrid assays, Scott Michaels and Laurel Bender for performing the unbiased Emc Y2H screen, and Rudi Turner for the SEM images. We also thank Spyros Artavanis-Tsakonas, Antonio Baonza, Nick Baker, Brian Calvi, Robert Holmgren, Yuh-Nung Jan, Gerald Rubin, Adi Salzberg, the Bloomington Drosophila Stock Center (BDSC) and the Drosophila Genomics Resource Center (DGRC) for fly stocks, antibodies and DNA plasmids. Finally, we thank Utpal Banerjee for generously allowing C.M.S. to complete some of the work for this manuscript in his lab.