The Notch pathway regulates the Second Mitotic Wave cell cycle independently of bHLH proteins

Abstract

Notch regulates both neurogenesis and cell cycle activity to coordinate precursor cell generation in the differentiating Drosophila eye. Mosaic analysis with mitotic clones mutant for Notch components was used to identify the pathway of Notch signaling that regulates the cell cycle in the Second Mitotic Wave. Although S phase entry depends on Notch signaling and on the transcription factor Su(H), the transcriptional co-activator Mam and the bHLH repressor genes of the E(spl)-Complex were not essential, although these are Su(H) coactivators and targets during the regulation of neurogenesis. The Second Mitotic Wave showed little dependence on ubiquitin ligases neuralized or mindbomb, and although the ligand Delta is required non-autonomously, partial cell cycle activity occurred in the absence of known Notch ligands. We found that myc was not essential for the Second Mitotic Wave. The Second Mitotic Wave did not require the HLH protein Extra macrochaetae, and the bHLH protein Daughterless was required only cell-nonautonomously. Similar cell cycle phenotypes for Daughterless and Atonal were consistent with requirement for neuronal differentiation to stimulate Delta expression, affecting Notch activity in the Second Mitotic Wave indirectly. Therefore Notch signaling acts to regulate the Second Mitotic Wave without activating bHLH gene targets.

INTRODUCTION

The organization of growth and of cell division is arguably as important for proper development as is the patterning and differentiation of cell types. The patterning of cell division may be as well understood in the Drosophila eye as in any other tissue. Proliferation in the differentiating eye disc ensures generation of adequate progenitor cells for the precise, repetitive organization of the retina. The cell-cell signaling events that associate cell cycle arrest with neuronal differentiation and coordinate the proliferation of progenitor cells have all been characterized, both in normal development and in regeneration (Baker, 2007; Fan and Bergmann, 2008a; Fan et al., 2014)..

Drosophila eye differentiation begins in the early third larval instar, when a wave of differentiation initiates at the posterior margin of the eye imaginal disc, traveling anteriorly until the entire organ is differentiating (Ready et al., 1976; Treisman, 2013). The wave of differentiation is preceded by wave of G1 phase cell cycle arrest, during which the first cell types are specified and begin to differentiate. Subsequently, the remaining retinal progenitor cells (the majority) are born in a ‘Second Mitotic Wave’ cell cycle that occurs just posterior to the morphogenetic furrow, after the first photoreceptor cell types have begun to differentiate (Figure 1A–D)(Ready et al., 1976; Baker, 2001).

Figure 1. The Second Mitotic Wave and Notch signaling.

All panels show eye imaginal discs labeled with antibodies against Cyclin B (red), the neuronal differentiation marker Elav (blue), and either the mitotic antigen phospho-Histone H3 (white in panel A) or the genotype marker GFP (panels E,F,I,J). A–D, wild type eye disc. The pattern of mitosis (anti-pH3: panels A, B) shows a Second Mitotic Wave (arrowhead) that follows a wave of cell cycle quiescence. Cyclin B accumulation (panels A, C) shows S-phase entry and cell cycle progression by the SMW cells while the first photoreceptor cells are beginning terminal differentiation (Elav, panels A, D). Insert in panel D shows the normal photoreceptor differentiation pattern enlarged. pH3 (panel B) and Cyclin B (panel C) overlap in early metaphase cells where Cyclin B surrounds the condensed, pH3-containing mitotic chromosomes (green arrows indicate examples). E–H, Su(H) mutant eye disc. Flp-mediated recombination has produced an eye disc largely homozygous for Su(H) with only a few remaining Su(H)/+ cells (marked by GFP, panel F, green in panel E). No SMW Cyclin B activity is apparent (panel G). Almost all cells differentiate as neurons (panel H; insert shows enlargement) Genotype: eyF; Su(H)^D FRT40A/M(2)24 FRT40A p{ubi-GFP}. I–L. mam mutant eye disc. Flp-mediated recombination has produced an eye disc largely homozygous for mam with only a few remaining mam/+ cells (marked by GFP, panel J, green in panel I). CycB reports many cells cycling at the SMW stage (panel K). Almost all cells differentiate as neurons, although some cells may delay differentiation slightly in comparison to Su(H) (panel L; insert shows enlargement). Genotype: eyF; FRT42D mam¹⁰/FRT42D ubiGFP M(2)56f.

Hedgehog and Dpp signaling contribute to the arrest of the cell cycle anterior to the morphogenetic furrow (Horsfield et al., 1998; Firth and Baker, 2005). Because Hh and Dpp are also required to initiate differentiation (Greenwood and Struhl, 1999; Curtiss and Mlodzik, 2000), differentiation and cell cycle arrest may be coordinated by co-dependence on these shared signals.

Once differentiation has begun, entry into the SMW occurs in the remaining undifferentiated cells, and is regulated by Notch and by the EGF-receptor, acting in opposite directions. Notch promotes S-phase entry into the SMW after the first retinal photoreceptors have already begun differentiation, whereas EGF-receptor signaling blocks cell cycle entry in coordination with the onset of neuronal differentiation (Baker and Yu, 2001; Yang and Baker, 2003; Baonza and Freeman, 2005; Firth and Baker, 2005). The EGF receptor pathway also plays a role later in the cell cycle of retinal progenitor cells, where it is required for G2-M progression of the SMW cells. Most cells do go on to mitosis, although about 11% remain in G2 until the pupal stage, when they usually either undergo further divisions to differentiation as the inter-ommatidial bristle organs, or die (Baker, 2001; Meserve and Duronio, 2017). Other SMW cells normally remain in G1 before differentiating into various retinal cell types, but they can be forced into ectopic cell cycles by growth signals such as Cyclin D/Cdk4 or Insulin/IGF-1 receptor over-expression, or ectopic Cyclin E(Richardson et al., 1995; Baker, 2013). Compensatory cell division that follows damage depends on Hh and EGFR signaling (Fan and Bergmann, 2008b; Fan et al., 2014).

Prior to their implication in cell cycle control, Notch and the EGF receptor were already known to regulate patterning of eye differentiation. The EGF-receptor is required for the R1–R7 photoreceptor cell fates (as well as other steps in eye development)(Baker and Rubin, 1989; Xu and Rubin, 1993; Freeman, 1996; Treisman, 2013). Notch signaling suppresses photoreceptor differentiation, so that loss of Notch function leads to a ‘neurogenic’ phenotype of excess neural differentiation, also seen in many other tissues when Notch pathway function is removed (Cagan and Ready, 1989; Artavanis-Tsakonas et al., 1995). A potential consequence of control of differentiation and proliferation by the same pathways is to facilitate coordination of cell cycle and differentiation events. For example, Notch signaling may coordinate cell cycle entry with maintenance of an unspecified state, since it controls both processes (Yang and Baker, 2006). While the signal transduction pathway and immediate transcriptional targets by which Notch signaling regulates neuronal differentiation in the retina is known (Ligoxygakis et al., 1998b; Bray, 2006; Treisman, 2013), the cell cycle targets have not been identified

The signaling pathways acting in Drosophila eye development also regulate cell proliferation in many other circumstances. These pathways can be oncopgenic in mammals. Whereas Cyclin D/Cdk4 and Insulin/IGF-1 receptors are directly linked to cellular growth and therefore associated with positive cell cycle signals (Choi and Anders, 2014), Notch, EGF receptor and Hh are versatile developmental signals whose specific effects on cells in particular developmental contexts varies according to the state of the cells (Andersson et al., 2011; Zeng and Harris, 2014; Pak and Segal, 2016). For example, at an earlier stage of Drosophila eye development, EGF receptor is required for general growth of the tissue, opposite to its later role inhibiting S-phase entry in the SMW(Dominguez et al., 1998).

It is thought that a transcriptional target of Notch signaling is involved in SMW cell cycle entry since this depends on Su(H), the transcription factor activated by Notch signaling (Firth and Baker, 2005). Cyclin E is a common transcriptional regulator of S-phase entry in Drosophila, but Cyclin E protein was reported to accumulate at higher levels in the differentiating photoreceptor cells that do not enter the SMW than in the SMW cells, suggesting that it might not be cyclin E that was limiting for entry into this cell cycle (Firth and Baker, 2005). It was suggested that Cyclin A might have a role in cell cycle entry in the SMW(Baonza and Freeman, 2005). Genes with transcription patterns that correlate with the SMW have been identified (Firth and Baker, 2007), but only a small subset of such genes are cell cycle regulatory genes (Dimova et al., 2003; Rustici et al., 2004). Recently, Emc was suggested to be the target of Notch that regulates the SMW(Spratford and Kumar, 2015). Emc is the Drosophila Inhibitor of DNA binding (Id) protein homolog and a negative regulator of bHLH-mediated differentiation (Campuzano, 2001). Emc contributes to cell proliferation in the wing disc and undifferentiated eye disc by restricting the activity and expression of the bHLH protein Daughterless (Bhattacharya and Baker, 2011; Wang and Baker, 2015). Another potential target could be myc, which is an important target of mammalian Notch during leukemogenesis (Weng et al., 2006). Consistent with this, cells entering the SMW appear to show a transient increase in levels of the nucleolar protein fibrillarin, which is a myc-target gene in Drosophila (Grewal et al., 2005; Baker, 2013).

Here, we explore the roles of multiple components of the Notch signal transduction pathway in regulation of the SMW, reporting multiple differences in comparison with the regulation of photoreceptor differentiation. These differences might prove useful in identifying cell cycle Notch targets, since we present and discuss evidence against the hypotheses that any of Cyclin A, Emc, or Myc are the Notch targets regulating SMW entry.

RESULTS

Previous studies show that the Second Mitotic Wave is cell-autonomously absent in clones of cells lacking either Notch or its main transcription factor effector Su(H)(Baonza and Freeman, 2005; Firth and Baker, 2005). In such clones, all cells posterior to the morphogenetic furrow remain in G1, and many differentiate as neurons because of the roles of N signaling repressing the differentiation of photoreceptor neurons (Cagan and Ready, 1989; Treisman, 2013). This is illustrated in Figure 1E–H, which shows an eye disc in which nearly all cells are homozygous for a Su(H) null mutation. This eye disc lacks a Second Mitotic Wave, as indicated by the absence of Cyclin B protein that would normally accumulate in the SMW cells. Cyclin B is a reliable cell cycle reporter, used in many prior studies including those that established this cell-cycle role of Notch, because degradation of Cyclin B protein by APC/C between mitotic metaphase and the G1/S transition is a fundamental feature of cell cycle progression. Because of this, Cyclin B protein accumulation can only begin in cells that have entered S-phase, and Cyclin B disappears at metaphase (Evans et al., 1983; Acquaviva and Pines, 2006). Accordingly, activity of the Cyclin B degron is used to indicate both S-phase entry and passage through mitosis in the FLY-FUCCI (fluorescent ubiquitination based cell-cycle indicator) system (Zielke et al., 2014). An advantage of monitoring Cyclin B protein itself over FUCCI, in addition to greater simplicity of genetic strains, is the subcellular localization of Cyclin B protein, which is cytoplasmic until mitotic prophase when the nuclear envelope breaks down, so that early metaphase cells can be recognized from their nuclear Cyclin B accumulation (eg Figure 1A–D).

The Su(H) mutant eye discs also exhibited a strong neurogenic phenotype ie increase in neuronal differentiation, reflecting the well-known role of the Notch pathway in restraining photoreceptor differentiation in the eye disc, similar to the role of Notch in inhibiting neural fate determination in most tissues (Fortini and Artavanis-Tsakonas, 1994; Bailey and Posakony, 1995; Schweisguth, 1995; Ligoxygakis et al., 1998a). It is not thought, however, that lack of the SMW in Su(H) mutant clones is simply due to prior diversion of all progenitor cells to neural fate, because the SMW is also absent from Su(H) Mad ci mutant clones in which neurogenesis is diminished considerably (Firth and Baker, 2005).

A similar experiment performed with the Su(H) coactivator mam led to a different result (Figure 1I–L). Many cells entered a Second Mitotic Wave in eye discs that almost completely lacked mam function. This mam mutant cell cycle entry showed similar timing to that seen in wild type, but it was not normal spatially, lacking the regular hexagonal array that normally surrounds the permanently-arrested photoreceptor neurons of the ommatidial preclusters (Figure 1I–L). The mam null mutant cells also exhibited a strong neurogenic phenotype and the null clones contained many excess photoreceptors (Figure 1I,L). Since mam was required for the patterning of eye neurogenesis, it was difficult to determine whether mam mutant eye discs exhibited a different cell cycle pattern because of the altered pattern of neurogenesis, and alterations to the level and pattern of Delta expression that is expected to accompany this, or because mam also affected cell cycle entry to some degree. What was clear, however, is that many cells could enter the SMW cell cycle in the absence of mam.

The main transcriptional targets of Notch signaling during neurogenesis are the transcriptional repressor bHLH genes of the E(spl)-C, and their transcription requires both Su(H) and Mam (Fortini, 2009; Yamamoto and Bellen, 2014). Clones of E(spl) null cells were examined to see whether the cell cycle was affected. Little SMW cell cycle entry was seen in E(spl) mutant clones in initial experiments, similar to the result in the absence of Su(H) and different to that seen for mam (Figure 2A–E). Because E(spl) clones show a strong neurogenic phenotype in the eye, like other Notch pathway mutants (Figure 2C)(Ligoxygakis et al., 1998a), it was important to determine whether the absence of cycling cells was due to failure of cell cycle entry by eye disc precursors in the absence of E(spl), or to absence of any competent precursors when all of the mutant cells having been transformed to photoreceptors, as this would preclude their cell cycle entry. Clones of cells double-mutant for E(Spl) and for atonal (ato) were examined to investigate this. The proneural gene ato is required for specification of the R8 photoreceptors that found each ommatidium (Jarman et al., 1994). In the absence of R8, other photoreceptors are not specified either, and all cells are available to enter the SMW(Figure 2F–J)(Jarman et al., 1994). Cell cycle entry in ato single mutant clones has been described before (Baker and Yu, 2001). Many cells re-enter the cell cycle posterior to the morphogenetic furrow in a patchy fashion, whereas others do not. Most cells that re-enter the cell cycle are close to the posterior boundary of ato clones (Figure 2F,I,J). This distribution may be explained by non-autonomous rescue of cell cycle entry only near the wild type cells that produce normal levels of Dl (see Discussion). Dl is predominantly produced by photoreceptor cells posterior to the morphogenetic furrow, and Dl expression is greatly reduced in ato mutants (Parks et al., 1995; Baker and Yu, 1998). This would be expected to reduce Notch activity within ato mutant clones except within range of the clone boundaries.

Figure 2. The Second Mitotic Wave in the absence of E(spl).

Cell cycle progression in cell clones that lack GFP (green, and panels B,G,L,Q). A E(spl) null clone. Enlarged in panels B–E. E(spl) null clones (arrowheads) exhibit a strong neurogenic phenotype (Elav labeling, panel C and blue in panels A,E) and lack the SMW as indicated by Cyclin B labeling (panel D and red in panels A,E). F. Clones lacking ato function, enlarged in panels G–J. ato clones (arrowhead) lack neuronal differentiation (Elav labeling, panel H, and blue in panels F,J) and show entry into the SMW only near the clone borders (indicated by Cyclin B accumulation, panel I and red in panels F,J). K Double mutant clones lacking ato and E(spl), enlarged in panels L–O. The ato E(spl) clones (arrowheads) lack neuronal differentiation (panel M, and blue in panels K,O) and still exhibit entry into the SMW near the clone borders (panel N, and red in panels K,O). P. Double mutant clone lacking ato and E(spl), enlarged in panels Q–S. ato E(spl) cells can undergo mitosis near clone borders (arrows indicate phospho-Histone H3 labeled mitotic figures in panel R and magenta in panels P,S).

Cell cycle entry in ato E(spl) clones was largely similar to that seen in ato clones (Figure 2K–O). That is, cell cycle progression occurred non-autonomously near the posterior edges of ato E(spl) clones, as measured by Cyclin B accumulation, and mitotic figures were also observed within ato E(spl) clones, as revealed by labeling for phospho-Histone H3(Figure 2P–S). These findings suggested that cell cycle regulation by Notch could occur independently of E(spl), when differentiation was blocked due to the absence of ato. Like mam, it was difficult to determine whether cell cycle entry was completely independent from E(spl), because of the changes in neurogenesis and patterning in such genotypes, but it was clear that substantial cell cycle entry could occur in the absence of both ato and E(spl).

To examine another aspect of the Notch pathway, the role of ubiquitin ligases was examined. Ubiquitinylation of the ligand Dl by Neur plays an important role in patterning eye neurogenesis (Weinmaster and Fischer, 2011; Kandachar and Roegiers, 2012). Clones of cells lacking neur showed a significant reduction in entry into the SMW(Figure 3A–D). As discussed for E(spl), this could be an indirect effect of the neurogenic phenotype of neur mutant clones reducing the number of uncommitted progenitor cells available for cell cycle entry (Figure 3A,C). To reduce this indirect effect, clones of cells double mutant for neur and pnt were examined. The Pnt transcription factors mediate differentiation in response to signaling by the receptor tyrosine kinases EGF receptor and Sevenless receptor (Treisman, 2013). Because of this, more cells enter the SMW cell cycle in the absence of pnt, where most of them arrest in G2 because EGFR-dependent G2/M progression is reduced (Figure 3E–H)(Yang and Baker, 2003). The neur pnt mutant cells showed greatly reduced neurogenesis, like pnt mutant cells, confirming that RTK signaling was epistatic to N-mediated lateral inhibition in the regulation of photoreceptor recruitment (Figure 3I–K). Greater cell cycle entry was observed in neur pnt clones than in neur clones, indicating that entry into the SMW could occur in the absence of neur activity, particularly if differentiation was suppressed (Figure 3I,L).

Figure 3. The Second Mitotic Wave in the absence of Neuralized and Mind-bomb.

Cell cycle progression is shown cell clones that lack beta-galactosidase (green, and panels B,F,J), and in homozygous mib1 mutants (M–O). A–D Clones homozygous for neur (arrrowhead) exhibit a strong neurogenic phenotype (Elav labeling in blue, and panel C) and lack the SMW as indicated by Cyclin B labeling (red, and panel D). E–H. Clones lacking pnt (arrowhead) lack neuronal differentiation except for R8 photoreceptors (Yang and Baker, 2003) (Elav labeling in blue, and panel G). Most cells enter the SMW but accumulate Cyclin B without undergoing mitosis (Cyclin B accumulation, red, and panel H). I–L Double mutant clones lacking neur and pnt (arrowheads) lack most neuronal differentiation (blue, and panel K) and many cells exhibit entry into the SMW with Cyclin B accumulation (red, and panel L). M–O. Eye disc from homozygous mib1 mutant. Cyclin B accumulation reports SMW cell cycle entry (red, and panel N), slightly anterior to the onset of neuronal differentiation (Elav, blue and panel O).

Dl can also be activated by a second ubiquitin ligase, Mindbomb, although Neur plays a major role during neurogenesis (Weinmaster and Fischer, 2011; Kandachar and Roegiers, 2012). SMW cell cycle entry occurred in eye discs mutant for mib (Figure 3M–O). Although it was difficult to assess whether this SMW was completely normal, since the mib mutant eye discs were smaller in size, it was evident that many cells could enter a SMW cell cycle in the absence of mib.

Incomplete requirements for ubiquitin ligases prompted further examination of the roles of Notch ligands. It was reported previously that Dl was required for SMW cell cycle entry, but the published data shows that while the SMW is strongly affected in clones lacking Dl, some cell cycle activity remained (Baonza and Freeman, 2005). To determine whether this could reflect partial redundancy with Serrate (Ser), the other Notch ligand, double mutant Dl Ser clones were examined here. Dl Ser mutant clones were generated in a Minute background, to increase the size of the mutant clones. The SMW was suppressed in Dl Ser mutant clones, but cell cycle entry was nevertheless detected in some cells (Figure 4A–H). Although cycling cells seemed more frequent close to the boundaries of Dl Ser mutant clones, where mutant cells could potentially be activated by ligands from nearby wild type cells, some cell cycle progression was seen in cells not close to wild type cells (Figure 4A–H). It is unlikely that Dl protein could persist in these large clones since before the mitotic recombination, because previous studies indicate that Dl protein is reduced to very low levels anterior to the morphogenetic furrow prior to then being re-expressed (see Figure 7H), so that even Dl mutant clones induced in the last cell cycles exhibit mutant phenotypes (Li et al., 2003; Miller et al., 2009). These results are generally similar to those reported previously for Dl single mutant cells (Baonza and Freeman, 2005).

Figure 4. The Second Mitotic Wave in the absence of Delta and Serrate.

Cell cycle progression (Cyclin B accumulation) is shown cell clones that lack Dl and Ser. In panels A–D, Dl Ser homozygous clones lack GFP (green, and panel D). The outline of the clone is shown in panels B,C. Dl Ser clones exhibit a neurogenic phenotype (Elav in blue and panel B). Although cell cycle activity in the SMW is reduced, some Dl Ser cells still enter the cell cycle (Arrows; Cyclin B in red, and panel C). Panels E–H show a second Dl Ser clone lacking GFP (green, and panel H). The clone shows a neurogenic phenotype (Elav in blue and panel F). In this clone some Cyclin B expression (arrow) is seen distant from the clone boundary (Cyclin B in red, and panel G). In panels J–M, Dl Ser homozygous MARCM clones express GFP and also the EGFR antagonist Argos (aos). Clones are positively marked (GFP: green and panel M). Clone outlines are shown in panels K,L. More neurons differentiate inside clones, though less than in plain Dl Ser clones (Elav in blue, and panel K). Some mutant cells enter the cell cycle, especially near clone borders (Cyclin B in red, and panel L). In panels N–Q, Dl Ser homozygous MARCM clones express GFP and also Ras^DN. Clones are positively marked (green, and panel F). Clone boundaries are indicated in panels O,P. Neurogenesis is increased inside clones, to a lesser degree than in plain Dl Ser clones (Elav in blue and panel O). Although the SMW is reduced, multiple cells enter the cell cycle (Cyclin B in red, and panel P). Panels R–S show an enlargement (3X) of the clone shown in panels N–P. Only apical confocal planes are projected for the Cyclin B and GFP panels (panels T,U) allowing individual cells to be resolved. At this very apical plane, GFP is cytoplasmic in most cells (nuclear GFP signal predominates in the projected panel Q). In addition to GFP-positive cells expressing Cyclin B that are in S-phase or G2-phase, two apical nuclei of cells entering mitosis are indicated with arrowheads.

Figure 7. The roles of emc, da, and ato in the SMW correlate with Dl expression.

A. Clones lacking emc were identified by the absence of beta-galactosidase labeling (green). Cell cycle progression revealed by Cyclin B (magenta, and see panel B). SMW entry occurred almost normally in emc clones (see particularly emc mutant region towards the top of the figure. Where clones extended posterior to the morphogenetic furrow, the SMW was spread out and displaced anteriorly by the greater speed of the morphogenetic furrow in such clones (eg arrow)(Bhattacharya and Baker, 2012; Spratford and Kumar, 2013). C. Clones lacking da were identified by the absence of beta-galactosidase labeling (green). SMW entry was largely absent from da clones (Cyclin B accumulation in green, and see panel D), although individual da mutant cells near the borders of clones could enter the SMW (eg arrows). E. Clones lacking da were identified by the absence of beta-galactosidase labeling (green, and see panel F). Mutant clones (eg white arrow) lacked most neuronal differentiation (Elav labeling in red, and see panel G) and showed reduced Dl protein levels posterior to the morphogenetic furrow (blue, and see panel H). I. Enlargement (2x) of the clone shown in E, with b-Gal channel (J), Elav channel (K) and Delta channel (L). Near the borders of clones, wild type R8 cells could recruit adjacent da mutant cells to neuronal differentiation and Dl expression, eg yellow arrows. These cells are seen as Elav-positive, bGal-negative cels near clones borders (eg yellow arrows). M. Clones lacking ato were identified by the absence of beta-galactosidase labeling (green, and see panel N). Mutant clones (eg white arrow) lacked most neuronal differentiation (Elav labeling in red, and see panel O) and showed reduced Dl protein levels posterior to the morphogenetic furrow (blue, and see panel P). Near the borders of clones, wild type R8 cells could recruit adjacent ato mutant cells to neuronal differentiation, seen as Elav-positive, bGal-negative cells (eg yellow arrows).

Like the other Notch pathway mutants, Dl Ser mutant cells were neurogenic in phenotype (Figure 4A–B,E–F). Two approaches were taken to suppress neuronal differentiation by inhibiting EGFR activity, similar to that described above where pnt was also mutated in the neur clones to suppress photoreceptor R1–R7 recruitment. First, ectopic argos (aos) was expressed in Dl Ser mutant clones using a MARCM approach. The aos gene encodes a negative regulator of EGFR activity. Dl Ser mutant clones over-expressing aos were less neurogenic in phenotype (Figure 4J–K), but still contained cells entering the SMW cell cycle, usually close to the clone boundaries (Figure 4 J–M). Secondly, dominant-negative Ras was expressed in Dl Ser mutant clones using a MARCM approach. Function of ras gene is required for recruitment of photoreceptors R1-7 to ommatidia, but Ras is not required for S-phase entry in the SMW, although like EGFR it is required for G2/M progression (Simon et al., 1991; Yang and Baker, 2001, 2003). Expression of dominant-negative Ras also appeared to ameliorate the neurogenic phenotype of Dl Ser clones although neurogenesis was still abnormal (reduced Ras signaling is not expected to affect differentiation of supernumerary R8 cells, which may be the major photoreceptor cell type in Dl Ser clones) (Figure 4N–O, R–S).. When the cell cycle was examined, it was clear that although the SMW was significantly reduced in Dl Ser clones expressing Ras-DN compared to the wild type, significant cell cycle entry occurred at a similar stage to the SMW in wild type regions of the eye disc, and some of these Cyclin B positive cells progressed through mitosis (Figure 4N–U). The dividing cells were undifferentiated as indicated by absence of Elav labeling (Figure 4N–U).

The requirement for Su(H) suggested that N signaling promoted cell cycle entry through a transcriptional target that has yet to be identified, because Su(H) encodes a transcription factor. One potential candidate was myc, since mammalian c-myc is a target of Notch signaling (Weng et al., 2006). Indeed, human Notch was first identified as a proto-oncogene that causes leukemia through c-Myc. To test the contribution of myc in the Drosophila eye, we made clones of cells homozygous for the null allele myc⁴. These clones were very small, indicating a role for myc in eye disc growth anterior to the morphogenetic furrow like that found for the growth of other imaginal discs (Figure 5A,B). It was clear, however, that myc⁴ homozygous cells could enter the morphogenetic furrow with normal timing and pattern, indicating that myc is not essential for the SMW and cannot be responsible for its regulation by Notch signaling (Figure 5A–D).

Figure 5. The Second Mitotic Wave in the absence of Myc.

A. Clones homozygous for a myc null allele lacking beta-Galactosidase labeling (green, and see panel B) only grow to small sizes in eye discs (yellow arrows). Despite this effect on growth, cell cycle entry occurs normally in the SMW (Cyclin B labeling in red, and see panel C). Individual myc mutant mitotic cells are seen in the SMW (magenta arrows). Mutations in myc have little effect on neuronal differentiation (Elav labeling in blue, and see panel D).

One transcriptional target of N signaling, in the eye as in other tissues, is the Helix-Loop-Helix gene emc (Baonza and Freeman, 2001; Adam and Montell, 2004; Bhattacharya and Baker, 2009). Emc protein is a general inhibitor of bHLH protein function, including both expression and function of the E-protein Daughterless. Most or all variation in Da levels in different tissues has been attributed to regulation by Emc levels (Bhattacharya and Baker, 2011). Recently, it was proposed that N regulates the eye cell cycle by upregulating emc, thereby inhibiting Da, which is proposed to inhibit the cell cycle, so that a barrier to cell cycle entry is removed in cells deficient for the Notch pathway (Spratford and Kumar, 2015).

Emc protein levels seem to be less affected by N signaling than the emc enhancer traps that have been the main focus of other cell cycle studies (Bhattacharya and Baker, 2009; Spratford and Kumar, 2015). In particular, we detected little change in Emc or Da protein levels in clones mutant for Su(H) (this was reported as data not shown in a previous publication and is shown directly here)(Bhattacharya and Baker, 2011)(Figure 6A–H). Next the cell cycle was examined in cells mutant for emc. Clones of emc mutant cells showed accelerated morphogenetic furrow movement as described previously (Brown et al., 1995; Bhattacharya and Baker, 2009, 2012) but aside from this faster development the Second Mitotic Wave cell cycle appeared relatively normal (Bhattacharya and Baker, 2009, 2011)(Figure 7A,B).

Figure 6. Expression of Emc and Da is largely independent of Su(H).

A. Clones lacking Su(H) were identified by the absence of GFP labeling (green, and see panel C). Emc protein (red, and see panel C) is normally reduced in the morphogenetic furrow region (arrowhead) (Brown et al., 1995), and this is not affected in Su(H) clones. Downregulation of Emc protein was not much affected by loss of Su(H). This particular clone is too anterior to exhibit a defect in neuronal differentiation (Elav labeling in blue, and see panel D). E. Clones lacking Su(H) were identified by the absence of GFP labeling (green, and see panel F). Da protein (red, and see panel G) is normally elevated in the morphogenetic furrow region (arrowhead) (Brown et al., 1996). Da protein levels were not much affected by loss of Su(H). Posterior to the furrow, Su(H) clones show strong a neurogenic phenotype (Elav labeling in blue, and see panel H).

Finally, the proposition that da inhibits the cell cycle was examined. This proposition contradicts previous findings that da is required for the SMW(Brown et al., 1996). Our experiments confirmed the original conclusion, that cell cycle entry was severely impaired in da mutant clones (Figure 7C,D). The cell cycle defect of da mutant clones was not completely cell-autonomous, and some cells near the borders of clones entered the cell cycle (Figure 7C,D). Da is the obligate heterodimer partner of Atonal that is required for ato function during eye development (Brown et al., 1996). Since the range of non-autonomous cell cycle rescue seemed reduced in da clones compared to ato (compare Figure 2F–J with Figure 7C,D), this may indicate a function for da that is independent of ato. It is already known that while ato is essential for the differentiation of R8 photoreceptor cells, da is also required for R2,3,4,5 photoreceptor cells (Jarman et al., 1994; Brown et al., 1996). This additional role for da in R2,3,4,5 differentiation might be relevant because differentiating photoreceptor precursors are the main source of Dl expression at this stage (Baker and Zitron, 1995; Parks et al., 1995; Tsuda et al., 2002). Lack of photoreceptor cells would be expected to reduce the source of the Dl ligand. Dl expression is greatly reduced in ato mutant eye discs, which may account for the non-autonomous reduction in the SMW in ato clones (Baker and Yu, 1998). As expected, Dl levels were also reduced in ato or da mutant clones (Figure 7E–P). We confirmed that both photoreceptor differentiation, and Dl expression, could occur near the boundaries of ato clones or da clones, where non-R8 cell differentiation persisted (Figure 7E–P), consistent with the non-autonomous cell cycle defects of ato clones and da clones (Figure 2F–J and Figure 7C,D).

DISCUSSION

Here we investigate how the Notch pathway regulates the cell cycle during Drosophila eye differentiation. Differences were observed between how Notch signaling regulates the SMW compared to how Notch regulates photoreceptor differentiation (summarized in Figure 8). The Notch pathway suppresses the specification of photoreceptor cells in a manner similar to regulation of neurogenesis by Notch in many other tissues (Ligoxygakis et al., 1998b). That is, activation of the transmembrane ligand Delta by the ubiquitin ligase Neuralized leads allows Delta to activate Notch, leading to release of the Notch intracellular domain, which then acts in a nuclear complex with Su(H) and Mastermind to induce the transcription of the E(spl)-C family of transcriptional repressors and prevent neural fate specification and differentiation (Fortini, 2009; Weinmaster and Fischer, 2011; Kandachar and Roegiers, 2012; Yamamoto and Bellen, 2014). By contrast to the pathway regulating neural differentiation, cell cycle entry in the SMW occurred in the absence of the E(spl)-C(Figure 2). This suggested that a distinct transcriptional target of Su(H) is involved, but unusually, this Su(H) function did not seem to require the co-activator Mam (Figure 1). Although most studies find that Mam is required for transcriptional activation by Su(H)(Janody and Treisman, 2011; Kitagawa, 2016), mutations in mam often have weaker phenotypes than other neurogenic genes (Lehmann et al., 1983). It is possible that Mam protein might exhibit exceptionally strong perdurance, but there is also in vitro evidence for mam-independent transcriptional activation by Su(H)(Cave and Caudy, 2008).

Figure 8. Model of the cell cycle and cell fate specification in the morphogenetic furrow.

A cartoon summarizing some of the genetic regulation within the morphogenetic furrow of the eye imaginal disc. General proliferation associated with unpatterned growth of the eye disc is arrested anterior to the furrow by Hh and Dpp signaling. These signals also induce Ato expression. Ato heterodimerizes with Da to specify R8 photoreceptor cells. R8 (and other photoreceptor cells) are a source of Dl signaling and Notch activation, which signals through E(spl) to restrict specification of R8 cells through lateral inhibition, and also restrict specification of R2,3,4,5 photoreceptor cells that are also recruited in the morphogenetic furrow. Notch signaling has been implicated in emc transcription but has little effect on Emc protein levels. Emc is a competitive inhibitor of Ato/Da function. Finally, Delta and Notch promote entry of unspecified progenitor cells into the Second Mitotic Wave cell cycle, through an unidentified gene target distinct from the E(spl) complex and which may not be Cyclin E. The fact that both photoreceptor differentiation and the SMW cell cycle are regulated by Dl through N may help coordinate the restriction of cell cycle entry to unspecified cells only.

SMW entry also occurred in the absence of neur (Figure 3). In the case of neurogenesis, cells lacking neur show a weaker neurogenic phenotype than cells lacking other components of the pathway, both in the embryonic CNS and in the eye, which is thought to reflect the activities of two other ubiquitin ligases, mindbomb and mindbomb2(Lai and Rubin, 2001; Li and Baker, 2001; Weinmaster and Fischer, 2011; Kandachar and Roegiers, 2012). We did not find any effect on the SMW of mutated mib1, but we did not examine mib2, or cells lacking two or more of these possibly redundant ubiquitin ligases (Figure 3).

We also examined the roles of ligands in more detail. Baonza et al reported that SMW entry depended on Delta, which is also the ligand necessary for regulation of neurogenesis in the eye as in other parts of the nervous system, but although their figures show an obvious reduction in cell cycle entry in the clones of cells lacking Dl, some cell cycle entry still seems to occur (Baonza and Freeman, 2005). We repeated their experiments using cells lacking both Dl and the other known ligand Ser, with similar results ie cell cycle entry was obviously disrupted in clones lacking both Notch ligands, nevertheless some cells still entered the cell cycle (Figure 4). A potential complicating factor is that reducing Notch function typically results in excessive neurogenesis, raising the question of whether some Dl Ser mutant cells were prevented from entering the cell cycle by differentiation as photoreceptors. The excess neurogenesis of such clones was partially suppressed by expression of Argos or Ras-DN(Figure 4). Although there is clearly a considerable disruption of the SMW in the absence of N ligands, there is also clear evidence of cell cycle entry, occurring at approximately the same time as the normal SMW.

There are at least three possible explanations for these findings. First, it it possible that another ligand besides Dl or Ser activates N in the SMW. Secondly, it has been reported that Dl can signal across several cell diameters using filopodia (Cohen et al., 2010). During neurogenesis, rescue of Dl mutant clones appears to extend beyond cells immediately adjacent to wild type cells (Li and Baker, 2001). Signaling across many cell diameters would be required to account for the cell cycle progression seen in Dl Ser mutant clones (Figure 4A–H). Thirdly, Notch can be activated independently of ligands if delivered to certain cellular compartments (Baron, 2012). Recently it has been suggested that such ligand-independent activation is suppressed by cis-inactivation, in which ligands inhibit N signaling in the same cells, so that ligand-independent activation can be revealed in cells lacking Dl and Ser (Palmer et al., 2014). Cis-inactivation is potentially an important contributor to the patterning of Notch-mediated lateral inhibition during neurogenesis, but direct evidence of such a role has not yet been obtained. If ligand-independent signaling is occurring in Dl Ser clones, some other pathway must be active near the SMW to explain why ligand-independent signaling cell cycle entry was limited to cells at the same stage posterior to the furrow as those that undergo the normal, ligand-dependent SMW(Figure 4).

The cell cycle targets of Notch are of some interest. Many Drosophila cell cycles depend on regulated transcription of Cyclin E. Cyclin E is clearly required for the SMW(de Nooij and Hariharan, 1995; Secombe et al., 1998; Sukhanova and Du, 2008). Firth and Baker (Firth and Baker, 2005) suggested that Cyclin E might not be limiting for SMW cell cycle entry, however, because Cyclin E was seen to accumulate more in the differentiating photoreceptor precursors that do not enter the SMW. Alternatively, the lack of Cyclin E accumulation in SMW cells might reflect Cyclin E protein instability in S phase, with accumulation only in differentiating cells that can’t enter the cell cycle (Sukhanova and Du, 2008). A Cyclin E transcriptional reporter appears to accumulate in the same cells as the Cyclin E protein (Duman-Scheel et al., 2002), however, which is not indicative of post-translational regulation.

Baonza et al suggested that Cyclin A was part of the Notch dependent machinery promoting cell cycle entry in the SMW, although they concluded it could not be the only important Notch target (Baonza and Freeman, 2005). What they observed, however, was dependence of Cyclin A protein levels on Notch signaling that exactly paralleled the dependence of S-phase entry on Notch signaling in various genotypes. Like Cyclin B, Cyclin A protein is degraded by APC/Cyclosome activity during G1(Lukas et al., 1999; Blanco et al., 2000), so that Cyclin A protein cannot accumulate until the G1/S transition. Therefore these results are also expected if Cyclin A simply accumulates whenever the G1/S transition has occurred, and is not a direct target of Notch signaling.

In mammalian cells, c-Myc appears to be an important Notch target. Indeed, human Notch was first identified as a proto-oncogene that causes leukemia through c-Myc (Weng et al., 2006). c-Myc is an activator of the nucleolar protein fibrillarin, which transiently increases during the SMW(Grewal et al., 2005; Baker, 2013). Although Dmyc is important for growth in Drosophila, so that clones of myc null mutant cells in the eye disc are small, they nevertheless enter the SMW at the normal time, indicating that myc is unlikely to be the critical Notch target in the SMW(Figure 5).

Emc is a transcriptional target of N signaling in the eye as in other tissues (Baonza et al., 2000; Adam and Montell, 2004; Bhattacharya and Baker, 2009). Even though Emc protein levels seem to be less affected by N signaling than is emc RNA(Figure 6A–D), it has been suggested that N regulates the eye cell cycle by upregulating emc, thereby inhibiting Da, which is proposed to inhibit cell cycle entry (Spratford and Kumar, 2015). Contrary to this model, previous studies have reported that Da is in fact required for the SMW(Brown et al., 1996). We investigated this potential role for emc directly, finding no evidence that emc was required for the SMW(Figure 7A,B). Our studies confirmed previous findings that da is required for S phase entry into the SMW, although not apparently for cell cycle activity anterior to the morphogenetic furrow (Brown et al., 1996)(Figure 7C,D).

The cell-autonomy of da requirement in the cell cycle had not been determined previously. We found that da mutant cells sometimes enter the SMW at the posterior boundaries of da clones, where they were adjacent to wild type cells (Figure 7C,D). This suggested that da was required for the production of a short-range non-autonomous signal that is required for SMW entry, and could be provided to da mutant cells that were near to wild type cells.

It is instructive to compare the effect of da on the cell cycle to that of ato. Ato is the heterodimer partner of Da that is required for R8 specification (Jarman et al., 1994). Da is also required independently of ato to specify R2,R3,R4 and R5. Interestingly, this requirement in R2-5 seems to be incomplete, since some da mutant R2-5 cells do differentiate at reduced frequencies (Brown et al., 1996). Because of this requirement for da outside R8, R1-7 photoreceptor cells can differentiate near borders of ato mutant clones, but fewer photoreceptors are expected near the borders of da mutant clones, and mostly of the R1, R6 and R7 cell types. Proneural genes often promote Notch ligand expression, and accordingly ato is required for normal Dl expression during these stages of eye development (Hinz et al., 1994; Kunisch et al., 1994; Heitzler et al., 1996; Baker and Yu, 1998). Reduced Dl expression is a plausible explanation for the absence of the SMW within ato clones, and the more extensive requirement for da in photoreceptor differentiation and Dl expression may explain why non-autonomous rescue of the SMW is more pronounced near the borders of ato clones than near the borders of da clones.

Our current view of the SMW is summarized in Figure 8. Unpatterned proliferation of the eye imaginal disc is terminated ahead of the morphogenetic furrow by Hh and Dpp signaling. Hh and Dpp affect many aspects of eye development including expression of master regulators of retinal determination such as Eyeless, Teashirt, Dachshund, and Sine Oculis, as well as Homothorax (Curtiss and Mlodzik, 2000; Bessa et al., 2002; Firth and Baker, 2009). Cells posterior to the furrow can still respond to growth, because extra cell cycles are driven in unspecified cells by overexpression of CycD/cdk4, Inr, or myc, as well as by Cyclin E(Richardson et al., 1995; Baker, 2013). It is possible that Notch regulates cellular growth in the SMW, since there appears to be a small but discernible increase in nucleolar size at this stage (Baker, 2013), and progenitor cells are not noticeably smaller after dividing, indicating that growth accompanies the SMW cell division. The SMW, however, does not depend on myc. It appears to depend on a transcriptional target of Su(H) outside of the E(spl)-C of bHLH genes and that can be transcribed in the absence of Mam. The possibility of a function of Su(H)/Nicd other than transcription cannot be ruled out. The SMW clearly depends on CycE, but CycE may not be the N-dependent gene required for SMW entry, because it is not obviously elevated in SMW cells. Contrary to a recent proposal (Spratford and Kumar, 2015), the SMW does not depend on the HLH gene emc and is not inhibited by the bHLH protein Da. Instead da is required for entry into the SMW(Brown et al., 1996). The positive requirement for da is cell-nonautonomous, as is the requirement for its heterodimer partner ato, and consistent with the role of ato and da in promoting Dl expression during retinal differentiation, which is likely to account for the cell cycle defect. Surprisingly, the requirement for Dl in the SMW cell cycle was not absolute, either because an unidentified ligand exists, long-range cell-nonautonomy occurs, or Dl (and Ser) may contribute to cis-inactivation of N, so that some ligand-independent Notch signaling could occur in the unphysiological circumstance that Notch ligands are completely absent.

MATERIALS AND METHODS

Mosaic Analysis

Mosaic clones were obtained using FLP/FRT mediated mitotic recombination (Xu and Rubin, 1993). Larvae were subjected to heat shock for 1 hour at 37ºC at 60±12 hours after egg laying to induce expression of hsFLP, and dissected 72 hours after heat shock. To make “flip-on” clones, larvae were heat shocked for 30 minutes. All flies were maintained at 25ºC unless otherwise stated. For the experiments shown in Figure 1E–L, eye discs of almost entirely mutant genotypes were obtained using the eyFLP transgene (Newsome et al., 2000) to stimulate recombination continuosuly in a Minute background.

Drosophila Strains

Strains employed included: da³ (Cronmiller and Cline, 1987); emc^AP6 (Ellis et al., 1990); act>CD2>Gal4 (Pignoni and Zipursky, 1997); ato¹(Jarman et al. 1994); ato³(Jarman et al. 1995); Su(H)^D47 (Morel and Schweisguth, 2000); mam¹⁰, neur¹ (Lehmann et al., 1983); E(spl)gro^b32.2 p{gro⁺} (Heitzler et al., 1996); mib^EY09780 (Bellen et al., 2004); pnt^D88 (Scholz et al., 1993); Dl^rev10 (Haenlin et al., 1990); Ser^RX106 (Thomas et al., 1991); UAS-Ras^N17 (Lee et al., 1996); myc⁴ (Pierce et al., 2004);

Immunohistochemistry and image processing

Antibody staining was performed as previously described (Baker et al., 2014). The following primary antibodies were used: mouse anti-βGal (1:100, DSHB 40-1a), rabbit anti-βGal (Cappell), rat anti-ElaV(1:50, DSHB 7E8A10), mouse anti-Da (1:200)(Cronmiller and Cummings, 1993), rabbit anti-Emc (1:8000, a gift from Y. N. Jan), rabbit anti-Ato (1:50000)(Jarman et al., 1994), rat anti-GFP(1:1000, Nacalai Tesque GF090R), mouse anti-Dl (mAb202). Secondary antibodies conjugated with Cy2, Cy3 and Cy5 dyes (1:200) were from Jackson ImmunoResearch Laboratories. Images were recorded using Leica SP2 and Leica SP5 confocal microscopes, and processed using ImageJ64 and Adobe Photoshop CS5.