Early Eye Development: Specification and Determination
The compound eye of Drosophila melanogaster consists of hundreds of stereotypically organized unit eyes called ommatidia. The development of the eye begins in the fly embryo with the formation of the precursor cells of the presumptive eye-antennal imaginal disc. The disc expresses genes that belong to a so-called retinal determination network (RDN). The interplay between the genes of the RDN specifies and directs the development of the adult eye. The RDN includes highly conserved genes and integrate canonical signalling pathways, yet the outcome is totally distinctive from a typical vertebrate lens eye. The members of the RDN are highly interconnected forming complex, non-hierarchal loops controlling the transcription of one another. Knowledge gained from studying the RDN has helped us to better understand important developmental processes, such as cell specification and tissue growth. Astonishingly it has also demonstrated that eyes across species and phyla, albeit so different, share a common origin. In this chapter we will summarize the current knowledge of the RDN with its members and their interactions. In addition, we will briefly introduce the early specification steps of the eye-antennal imaginal disc in the embryo and later in the larvae.
KeywordsImaginal Disc Ventral Nerve Cord Morphogenetic Furrow Antennal Disc Retinal Determination Network Gene
The compound eyes of insects are typically composed of a large array of unit eyes termed ommatidia (Fig. 1a). The number of ommatidia and the size of the eyes are variable within the group of insects. The compound eye of the fruit fly Drosophila melanogaster is composed of 750–800 ommatidia, forming a highly stereotypically organized, virtually crystalline lattice. In turn, each ommatidium is composed of photoreceptor (PR), cone, and pigment cells (reviewed in Wolff and Ready 1993). Most adult structures develop from larval epithelial structures called imaginal discs. The adult compound eye of Drosophila originates from the eye-antennal imaginal disc. The eye-antennal disc develops into the adult eye, antenna, head capsule, and the ocelli, a group of extra-retinal photoreceptors (Kenyon et al. 2003). The discs' precursors are specified during embryogenesis and the imaginal discs keep proliferating throughout the three larval instar stages. During third instar, retinal differentiation is initiated in the eye disc from posterior to anterior by the dynamic progression of an epithelial groove called the morphogenetic furrow (MF). After metamorphosis, the pair of eye-antennal discs has transformed into the whole head capsule of the adult fly.
The eye disc is specified and determined during embryogenesis and larval stages by a genetic network called the retinal determination network (RDN) . One of the earliest genes expressed in the presumptive eye field is called eyeless (ey). Mildred Hoge described the ey mutant in Drosophila almost 100 years ago (Hoge 1915), and genetically mapped the gene causing this phenotype to the fourth chromosome of the fly. Much later, the ey gene was cloned and sequenced, leading to the astonishing observation that this gene is a homolog of the vertebrate Pax6 gene, which upon mutation causes a developmental syndrome of retina called aniridia in humans and a similar disorder caused by the small eyes mutation in mice (Quiring et al. 1994; Walther and Gruss 1991; Hill et al. 1991; Ton et al. 1991). Support for the homologous function was further provided by cross-phylum genetic experiments in which the mouse Pax6 gene was shown to be able to replace the Drosophila ey gene , since its targeted expression in Drosophila results in the formation of ectopic eyes (Halder et al. 1995). While the evolutionary origin of eyes was still widely subject of debates in the field, this discovery challenged much of previous beliefs in the independent convergent evolution of various different eye types across species (Gehring 2002). The finding that both Drosophila and vertebrate genes share the same function in governing the formation of eyes strongly indicated that these organs have evolved from a common ancestral prototypic eye and therefore supports the theory of a monophyletic origin of the eye (Gehring 2002; Halder et al. 1995).
Moreover, in addition to ey, other members of the RDN that specify the eye field in Drosophila are homologs to the corresponding genes in vertebrates (reviewed in Wawersik and Maas 2000). Thus, even though the camera-type eye of vertebrates and the compound eye of Drosophila are morphologically very distinct, the molecular mechanisms underlying the early specification of an eye field are surprisingly conserved. This discovery has turned the Drosophila eye-antennal disc into an excellent model system to analyze the formation of vertebrate eyes and to model human diseases. Also, studying the specification of an eye field can teach us how early determination genes integrate multiple signaling pathways during the course of development.
In this chapter we will focus on the RDN, the genetic network underlying eye field specification and determination. In particular each of the members of the RDN and their interactions with each other will be discussed in detail. Further attention is paid to the development of the eye precursor cells during embryogenesis, and the establishment of a distinct eye field in the developing eye-antennal disc during the first and second instar stages. Finally, we will briefly summarize the knowledge gained so far on how some of the members of the RDN contribute to the specification of the extraretinal photoreceptors in Drosophila.
The Retinal Determination Network
Development of the eye field requires three distinct phases: specification, determination, and differentiation. A highly interconnected regulatory network called the retinal determination network (RDN) is responsible for the initial specification and determination steps (Fig. 2). This network consists of selector genes that are regulated by extracellular signaling pathways and interact in highly complex regulatory and autoregulatory feedback loops. Thus, apart from the initiation of RDN gene expression , this network does not form a linear hierarchal gene regulatory system. Many of the RDN genes are not exclusively functioning in eye development, but are also involved in the formation of other tissues. Hence, it is the combinatorial effect and the interaction of these genes with each other that serve as a base to build an eye. Since most RDN genes also provide essential functions during embryogenesis, eye-specific mutant phenotypes are mostly the result of deletions in a regulatory region of an enhancer that is necessary to drive reporter expression only in the eye imaginal disc (Bui et al. 2000; Zimmerman et al. 2000).
Which genes are parts of the RDN and what are their characteristics? While there is no clear definition of a RDN gene, the network members have to fulfill certain criteria. Clearly, these genes have to be active at some point between the formation of the eye disc and beginning of the differentiation of the retinal cells. Most of the genes are able to induce ectopic eyes when overexpressed in other imaginal discs, though they do not all have the same potential in doing so. Conversely, mutational inactivation of RDN genes should interfere with the development of the eye (Fig. 1b). Most of the RDN members are also nuclear proteins directly controlling or affecting transcription. Obviously, there are more factors than just the RDN genes involved in eye development; however, they do not fill the criteria to be considered as selector genes.
To current knowledge and applying the definition above, the RDN consists of following factors: the homologs of the vertebrate Pax6 eyeless (ey) and twin-of-eyeless (toy) , the six family members sine oculis (so) and optix, the tyrosine phosphatase Eyes absent (Eya), a winged helix containing transcription factor Dachshund (Dac), zinc-finger transcription factors Teashirt (Tsh)and Tiptop (Tio), the nuclear factors Eyegone (Eyg) and Twin-of-eyegone (Toe), homeobox containing transcription co-factor Homothorax (Hth), serine-threonine kinase Nemo (Nmo), and the Pipsqueak motif containing genes distal antenna (dan) and distal antenna related (danr). The RDN also requires five canonical extracellular signaling pathways: the Notch, epidermal growth factor receptor (EGFR), Hedgehog(hh) Transforming growth factor beta (TGFβ), and Wnt signaling pathways. The factors of these signaling pathways will be presented in context and will not be described in detail.
Players of the Retinal Determination Network
eyeless and twin-of-eyeless
Pax6 genes encode transcription factors containing two DNA binding domains; a 128 amino acid long paired domain and a homeodomain (Quiring et al. 1994; Treisman et al. 1991; Walther and Gruss 1991). Other Pax family members have a varying number of paired domains with homeodomains and all of them have important functions in nervous system development across different species (for references see Treisman et al. 1991). The paired domain can be divided in two subdomains, the N-terminal PAI and C-terminal RED. Both of them contain a helix-turn-helix (HTH) motif (for references see Jun et al. 1998).
Drosophila has two Pax6 homologs; ey and twin-of-eyeless (toy) . Ey and Toy share 95% and 91% amino acid sequence identity, respectively, in the paired domain and 90% in their homeodomains with the murine Pax6 protein (Quiring et al. 1994; Czerny et al. 1999). Toy in comparison to Ey shares more sequence similarities to Pax6 outside these domains and has an additional transcription activation C-terminal domain that is present in Pax6 but missing in Ey. In addition, Ey has one amino acid substitution (Asn to Gly) in the paired domain, which affects its binding affinity to some Pax6/toy binding sites and thus explains the differential expression patterns and functions of ey and toy in the fly (Czerny et al. 1999).
Both toy and ey arose most likely due to a gene duplication event during late insect evolution, since two Pax6-like genes are only present in holometabolous insects (Czerny et al. 1999). Originally, these two genes were acting in parallel after the duplication event. ey eventually gained a mutation affecting the amino acid composition of the paired domain and radically changing the mode of action of the protein. The changes in the protein sequence abolished the capability of ey) for autoregulation. toy and Pax6 among other species still posses this character. (Plaza et al. 1993). Since toy shares more sequence similarities with the vertebrate Pax6 gene and is essential for the head formation both in vertebrates and in the fly, it has been postulated that toy has been under stronger selective pressure than ey (Czerny et al. 1999).
Whole mount in situ hybridization experiments have shown that toy is the first RDN gene expressed in the embryo in areas giving rise to the eye precursors. The first transcripts of toy can be detected early in the posterior procephalic region of a stage 5 embryo (cellular blastoderm). After gastrulation, during stages 14–15, the expression pattern covers the optic lobe primordia, brain, parts of the ventral nerve cord (VNC), and eye-antennal imaginal disc anlagen. In the third instar larva toy expression is localized to both posterior and anterior of the MF and in the ocellar precursors (Czerny et al. 1999; Jacobsson et al. 2009).
The expression of ey first appears in the brain and VNC during germ band extension in the stage 9 embryo (Quiring et al. 1994; Czerny et al. 1999; Kammermeier et al. 2001). The ey transcripts can be localized earlier in the VNC, but later in the brain than those of toy. In addition, the ey expression pattern in the brain is more spatially restricted than that of toy. In the developing visual anlagen, ey expression is almost entirely overlapping with toy expression pattern (Czerny et al. 1999). Ey can also be detected in the mushroom bodies, optic lobes, and in the central complex of the adult brain, and different ey mutants show defects also in the development of these organs (Callaerts et al. 2001). These observations link Ey functionally closer to the vertebrate Pax6 which is crucial for the formation of certain brain areas (Callaerts et al. 2001 and references therein). The ey expression persists until the third instar larval stage and is detected similarly to toy in the anterior boundaries of the eye disc as well as in the regions anterior to the MF (Quiring et al. 1994; Czerny et al. 1999).
The ey loss-of-function mutants have different eye and head phenotypes depending on the severity of the mutation. In general, hypomorphic alleles of ey lead to eyeless flies, whereas in eye-specific null mutants Change this sentence to: the head is completely missing due to the lack of the whole eye-antennal disc(Halder et al. 1998; Kronhamn et al. 2002; Quiring et al. 1994). Eye-specific toy null and hypomorphic mutants (toy hdl and toy G7.39 ) are also mostly headless (Kronhamn et al. 2002; Punzo et al. 2002). Some escapers are able to form a head and even compound eyes, but the ocelli are always missing (Kronhamn et al. 2002). The phenotype of the toy hdl is the result of a truncation in the homeodomain, yet the transcript pattern of toy hdl corresponds to wild-type transcripts (Kronhamn et al. 2002).
Forced overexpression of ey using different enhancer lines can effectively induce ectopic eyes on the wings, legs, halters, and antenna (Halder et al. 1995), and none of the other RDN members have the ability to function as a potent ectopic eye inducer as ey. Targeted overexpression of toy can induce ectopic eyes in the wings, legs, and halteres (Czerny et al. 1999; Salzer and Kumar 2010).
sine oculis and optix
so and optix are both part of the RDN since mutants of these genes show severe defects in eye development, and their forced expression is able to induce ectopic eyes (Cheyette et al. 1994; Pignoni et al. 1997; Seimiya and Gehring 2000; Serikaku and O’Tousa 1994; Weasner et al. 2007). These two genes are a homeodomain and a SIX domain containing transcription factors and members of the SIX (Sine oculis box) gene family (Cheyette et al. 1994; Oliver et al. 1995b). Vertebrates have six members in this family called SIX1 to SIX6 (Kawakami et al. 1996a; Kawakami et al. 1996b; Oliver et al. 1995a; Oliver et al. 1995b; Toy et al. 1998). The Drosophila so gene is most closely related to the murine Six1/Six2 genes. It was long believed to be the homolog of Six3, because it is the only gene of the SIX family involved in eye development in the mouse (Oliver et al. 1995a; Seimiya and Gehring 2000; Seo et al. 1999). However, later it became evident that optix instead is the homolog of Six3 (Seo et al. 1999; Toy et al. 1998). In addition to so and optix, Drosophila has one more gene belonging to this group but representing a different subclass, DSix4, which has no shown functions in the formation of the retina neither in the fly nor in vertebrates (Seo et al. 1999).
so is required not only for the development of the adult compound eye but also for the formation of other structures of the visual system including the larval eye , termed Bolwig’s organ, the ocelli and the optic ganglia in the brain (Cheyette et al. 1994). In the developing retina so is required for the initiation and propagation of the MF as well as for the formation of the PRs posterior to the furrow (Pignoni et al. 1997). The gene locus of optix is near the so locus, suggesting that these two genes went through a gene duplication event. optix is an orthologue of the mouse Six3, which also has specific functions during the eye development. Although these two genes share extensive sequence similarities even in the DNA binding homeodomain , they show distinct functions during the eye formation in Drosophila and are not redundant.
In the embryo, so is expressed in the optic lobe primordia anterior to the cephalic furrow. Later, so is detected also bilaterally at the segment boundaries. At stage 16, so expression is restricted to the anterior region of the head, including the Bolwig’s organ (Serikaku and O’Tousa 1994). The eye-specific so expression in the eye discs is activated during the late second larval stage, right before the MF initiation takes place in the third instar stage. so is expressed in a gradient starting at the anterior and decreasing towards the posterior end of the disc, and persists until the end of the larval stage (Cheyette et al. 1994; Kenyon et al. 2003). The expression of so is not restricted to the anterior site of the MF but is also expressed within and posterior to it. In addition to the eye-antennal disc, so expression is also present in the optic lobes and the Bolwig’s organ primordia of the embryo, and in the leg discs (Cheyette et al. 1994; Seimiya and Gehring 2000). The expression pattern of so during early larval stages, however, remains unclear. Depending on the lacZ reporter lines used, β-galactosidase expression is detected in the first instar (Wang and Sun 2012) or starting from mid-to-late second instar stage (Kenyon et al. 2003).
optix is expressed in a similar, but not identical pattern to so. optix transcripts can be detected early in the blastoderm embryo on its anterior end. During germ band extension, when so expression can already be detected in the optic lobe primordia,optix is still detectable only at the very anterior region of the embryo. During the second instar stage, optix is expressed in the whole eye disc, but in the third instar, contrasting so, it becomes restricted to an area anterior to the MF resembling ey and toy expression. optix is also present in the wing and haltere discs and in a part of the antennal disc (Seimiya and Gehring 2000).
so loss-of-function mutant animals show severe developmental defects in the retina, impairing cell proliferation and PR formation, leading to extensive cell death anterior to the MF (Cheyette et al. 1994; Pignoni et al. 1997; Serikaku and O’Tousa 1994). However, analysis of so 3 null mutant clones revealed that the cell death is a consequence and not a cause, since the clones initially show massive overproliferation (Pignoni et al. 1997). so 1 mutants are in most cases lacking compound eyes and ocelli, being the strongest non-lethal and eye-specific loss-of-function mutation. Other mutants cause less severe effects in the eye formation including a reduced eye size or rough eyes. Ocelli in these mutants are mostly depleted or reduced (Heitzler et al. 1993; Cheyette et al. 1994). so D (droplet, drl) is a dominant negative allele of so that is able to recruit an optix co-factor Optix binding protein (Obp) (Kenyon et al. 2005). This mutation causes a phenotype similar to so 1 mutation with the exception that ocelli are not reduced and the mutation is homozygous lethal (Heitzler et al. 1993). A loss-of-function mutation of optix has not been generated so far, leaving the phenotype of such a mutant under speculation.
The initial experiments to produce ectopic eyes by so overexpression failed, and it was assumed that So needs the co-activator Eya for its function in the eye. However, a screening combining upstream activation sequence (UAS)-so responder line with 219 distinct Gal4 driver lines revealed ectopic eye formation in four cases in the antennal part of the eye-antenna imaginal disc. This portion of the disc normally lacks Eya, suggesting that so is able to induce eye tissue without the presence of intrinsic Eya. Interestingly, So even induces eya expression in these ectopic eye tissues. It still remains to be cleared if this induction is direct or not (Weasner et al. 2007). Another Gal4 screen revealed so-induced ectopic eye tissue also in the head (Salzer and Kumar 2010). Nevertheless, compared to the ability of ey to induce eye tissue ectopically, so is a considerably less potent inducer.
Forced expression of optix in all imaginal discs leads to ectopic eye formation only in parts of the antennal disc, and to formation of extra ocelli (Seimiya and Gehring 2000). Using the same set of the 219 driver lines that were crossed with UAS-so , it was demonstrated that optix is also capable to induce ectopic eye tissue in the wing and haltere discs (Weasner et al. 2007). Interestingly, Optix does not require the presence of Ey to induce ectopic eyes (Seimiya and Gehring 2000), but optix is not able to induce ectopic eyes in so or eya mutant background (an observation of Seimiya and Gehring 2000).
Strikingly, only two of the 219 driver lines were able to induce ectopic eyes with both UAS-so and UAS-optix. Further analysis of the regulatory domains of so and optix indicate that although these genes do not considerably differ in their homeodomains and probably even bind the same target genes, they differ significantly in their SIX-domains and in the amino acid composition at the C-terminus. The SIX-domain is involved in the selection of a binding partner, thus with distinct partners So and Optix could also have different targets. Interestingly, the C-terminal end of Optix is involved in repression of retinal formation (Weasner et al. 2007) and one common binding partner shared by so and optix is the transcriptional co-repressor Groucho (Gro), suggesting that also so might function as a repressor of transcription (Silver et al. 2003; Kenyon et al. 2005).
Two genes interacting distinctively with so and optix are the so binding protein (sbp) and obp, respectively. The sbp expression is detected posterior to the MF where so but no optix is present. obp in contrast shows co-expression with both so and optix, but does not bind to so in a yeast-two-hybrid assay and has no effect on eye development when expressed in regions where only So is present. The exact functions of these two genes in combination with their binding partners are yet to be discovered (Kenyon et al. 2005). In addition, only so has been associated with eya (Seimiya and Gehring 2000).
Taken together, the existence of different factors that can bind so in distinct genetic environments suggests that so probably has a dynamic role in eye development, acting both as an activator and repressor of transcription. A similar kind of a mechanism has been demonstrated in vertebrates. Phosphatase activity of Eya changes the activity of Dach1, which is bound to Six1, to become a transcription co-activator (Li et al. 2003; reviewed in Dominguez and Casares 2005). The so binding sites have also been located in ey, hh, and lozenge (lz) genes (Pauli et al. 2005; Niimi et al. 1999). The lz as a target is of particular interest, since it is expressed and functions posterior to the MF and is involved in the first differentiation steps of the PRs, thus linking cell fate specification with differentiation.
The eya gene belongs to the phosphatase subgroup of the haloacid dehalogenase (HAD) family of transcription co-activators including or having one member in Drosophila (Eya) and four in vertebrates (Eya1–4) (Bonini et al. 1993; Hanson 2001; Tootle et al. 2003). The phosphatase activity is located at the C-terminus of the gene, which is also involved in interacting with So and Dac (Pignoni et al. 1997; Chen et al. 1997). The transactivator domain lays at the N-terminus, containing a proline-serine-threonine-rich region (Silver et al. 2003). The vertebrate Eya2 can rescue the eye-specific eya null mutant phenotype, causing eyeless flies and thus revealing a high degree of conservation in the eya gene sequence (Bonini et al. 1997). eya is involved in the inhibition of cell death and promotes cell specification in the eye imaginal disc anterior to the MF (Bonini et al. 1993). It was also demonstrated that eya, together with so, is required for the initiation and propagation of the MF, and for the development of the PRs posterior to the furrow (Pignoni et al. 1997).Since eya is able to produce ectopic eyes it belongs to the group of key factors in the RDN.
The eya transcription starts in the blastoderm embryo in the future head segments, but cannot be detected in the eye primordial cells at this stage. In the second instar stage eya shows expression in the posterior regions of the eye disc that decreases toward the anterior and middle parts of the disc. In the third instar stage, eya is expressed both in the anterior and posterior regions of the MF and in the region giving rise to the ocelli (Bonini et al. 1993).
The eya null mutants are embryonic lethal. Viable eye-specific eya 1 and eya 2 mutants display an eyeless phenotype, but develop all the other head structures (Bonini et al. 1993; Fig. 1b) The eye-specific mutant phenotype is the result of a deletion in a regulatory region of the eya enhancer (Bui et al. 2000; Zimmerman et al. 2000). Other less specific eya mutants (eya 3cs and eya 4 ) lack, in addition to the reduced eyes, also ocelli (Zimmerman et al. 2000). Strong loss-of-function eya clift1 mutant clones cause overgrowth of the disc followed by a massive cell death, as observed in so 3 mutants (Pignoni et al. 1997).
The eye-specific enhancer deleted in eya 1 and eya 2 is sufficient to recapitulate the eya expression pattern in the eye disc. This enhancer is not expressed in ey 2 mutant background, and is not affected in so 1 and dac 3 mutants (Bui et al. 2000). Ectopic expression of dac and ey, but not eya or so, is able to induce ectopic expression of this enhancer.
Overexpression of eya is able to cause ectopic eye formation in the antenna, and with more copies of UAS-eya, ectopic eyes form also in the legs and wings (Bonini et al. 1997). Using other imaginal disc-specific drivers eya can also initiate eye formation in the halters and in the head (Salzer and Kumar 2010).
dac was discovered in an enhancer trap screen that was conducted to find novel regulators of PR specification. The gene identified in this screen was named dachshund, since the mutants show a recessive short-leg phenotype (Mardon et al. 1994). dac is required for the progression of the MF and in the formation of PRs. It belongs to a gene superfamily containing the vertebrate dac homologs Dach1/2 and proto-oncogenes Ski/Sno of the Ski-family that are involved in transcriptional repression (Hammond et al. 1998). Dach genes in the vertebrates are also involved in eye and limb development (Hammond et al. 1998). The Drosophila Dac protein contains two conserved domains, a DNA binding, winged helix domain called DD1 and another domain called DD2. DD1 is essential for the function of Dac in the eye. DD2 is required to facilitate the function of DD1 and to interact with Eya, although neither of these functions is required for eye development (Pappu et al. 2005; Tavsanli et al. 2004).
dac is expressed in the third instar disc at the posterior edge anterior to the MF (Mardon et al. 1994). Posterior to the MF, dac is expressed in PRs R1, R6, and R7 and in cone cells. dac expression can be detected in other imaginal discs including leg, antenna, and wing as well as in the developing nervous system of the embryo and larval optic lobes.
dac mutants have severely reduced or absent eyes, they are not able to form PRs and the progression of the MF is impaired (Mardon et al. 1994). Overexpression of dac leads to eye tissue formation in the antennal and leg discs, confirming its place in the RDN (Shen and Mardon 1997).
The dac expression can be induced by ectopic expression of ey, eya, and eya/so complex, suggesting that these genes control dac activity also in the normal developmental conditions (Chen 1997). In addition, dac expression requires the presence of the extracellular signal Decapentaplegic (Dpp) (Chen et al. 1999).
teashirt and tiptop
Drosophila Tsh and Tio are nuclear proteins containing three and four zinc finger motifs, respectively (Laugier et al. 2005). Forced expression of these genes can induce ectopic eyes, and this is why these genes were included in the RDN in the first place (Rubin 1998). The tsh expression was initially identified in the trunk region of an early embryo regulating patterning together with homeotic genes. The embryonic role of tsh is to repress head development, and certain loss-of-function mutations lead to trunk-to-head transformations (Fasano et al. 1991; Roder et al. 1992; de Zulueta et al. 1994). tsh promotes the growth and specification of the dorsal part of the eye disc mainly during second instar larval stage. On the ventral half tsh suppresses eye specification together with hth (Singh et al. 2002). tio is a paralog of tsh, and has analogous functions in the eye formation as tsh. These two genes most likely evolved after a duplication event that occurred during the evolution of Drosophilidae (Laugier et al. 2005; Bessa et al. 2009).
tsh expression begins in the trunk at an early stage 6 embryo proceeding tio expression, which commences at the posterior region at stage 10. In the early embryo their expression pattern is not overlapping. Co-expression of the genes begins at stage 12 embryo in certain regions of the CNS and epidermis, and increases during the development, but they also maintain distinct expression patterns (Laugier et al. 2005). In the eye disc tsh expression begins already in the first instar stage and occupies the whole disc, thus overlapping with ey and hth expression (Singh et al. 2002). During the second larval stage, tsh expression retracts towards the anterior part of the disc, and by the late third instar larval stage tsh expression is restricted in the eye disc anterior to the MF, symmetrically on the dorsal and ventral halves of the disc (Pan and Rubin 1998; Singh et al. 2002). tio is co-expressed with tsh in the third instar larval eye disc (Bessa and Casares 2005).
Weak loss-of-function mutations of tsh do not show any phenotype in the fly eye. The tsh null mutants are embryonic lethal. To study the role of tsh in eye development, tsh null mutant clones were generated using an x-ray-induced mitotic recombination (Pan and Rubin 1998). These mutant clones in the eye disc do not display any abnormal phenotypes, suggesting that tsh has a redundant function in the eye disc.
Overexpression of tsh induces ectopic eyes in head tissues and antenna, and these extra eyes have nearly normal ommatidia (Pan and Rubin 1998). Interestingly, also the opposite has been demonstrated, where suppression of tsh actually promotes ectopic eye formation and forced expression suppresses eye development. This, however, only occurs at the ventral margin of the eye disc. In the dorsal part of the disc tsh promotes eye development. Thus, tsh plays a dual role in regulating the growth of the eye disc. The level of tsh expression might play a significant role in the decision whether the eye development is suppressed or promoted, since overexpression of more than one copy of tsh enhances the phenotypes (Singh et al. 2002).
The tio null mutants are viable and show no aberrant phenotype in the adult (Laugier et al. 2005). Nevertheless, as in the case of tsh, forced overexpression of tio causes ectopic eye formation in the head region (Bessa et al. 2009). tsh is also able to induce ey expression in the antennal disc when ectopically expressed. Reciprocally, ey expression is necessary for tsh expression (Singh et al. 2002). tsh might also function upstream of so, eya, and dac since these genes are ectopically expressed upon forced tsh expression (Pan and Rubin 1998).
Tsh and Tio negatively regulate themselves (negative autoregulatory feedback loop) and each other (Bessa et al. 2009). They can also compensate for each other’s function. For example, tio is able to rescue embryonic lethality of tsh mutants when expressed in the regions normally occupied by tsh (Bessa et al. 2009).
eyegone and twin-of-eyegone eyg and toe encode for nuclear proteins showing similarities to a splice isoform of the vertebrate Pax6 gene termed Pax6(5a), containing only one DNA-binding domain, the RED domain (and no PAI domain) (Jun et al. 1998). These genes have a different function during the retinal specification compared to the Pax6 homologs ey and toy. eyg is involved in the eye and salivary duct development. It positively regulates the growth of the eye disc, but also acts as a transcriptional repressor (Jang et al. 2003; Yao et al. 2008). toe acts as a repressor as well, but in somewhat different regions than eyg. The eyg possesses two, and toe only one repressor domain, which may account for their different functional outcomes (Yao et al. 2008).
eyg and toe are expressed in the embryo in the salivary gland placode (SGP), in the dorsal head, and in the segments of the trunk. Starting at embryonic stage 17, eyg and toe are expressed in the eye-antennal disc precursors in the same cells that are also expressing ey and toy (Czerny et al. 1999; Jones et al. 1998; Jun et al. 1998; Quiring et al. 1994; Yao et al. 2008). During the growth of the disc eyg and toe are expressed both in the antennal and eye part. In the eye disc they continue to be expressed anterior to the MF until the third instar stage. By then the expression is restricted to a small cluster of cells at the dorsal–ventral boundary (Yao et al. 2008). This expression pattern is different to that of ey and toy, which are expressed in a much larger area anterior to the furrow (Quiring et al. 1994; Czerny et al. 1999; Yao et al. 2008). eyg and toe are not expressed in equal levels; eyg contributes the clear majority to the total amount of transcript.
Weak loss-of-function mutants of eyg have reduced or missing eyes, whereas strong mutations fail to form a head and cannot hatch from their pupal cases (Jang et al. 2003). The eye discs of such mutants are highly reduced in size and show extensive cell death. However, these mutants cannot be rescued by overexpression of the apoptosis inhibitor p35, indicating that apoptosis is not the sole reason causing the eyg phenotype. toe alone does not seem to play an essential role in the eye since downregulation of toe transcripts by miRNA does not cause any obvious phenotype (Yao et al. 2008).
Forced expression of eyg is capable producing ectopic eyes on the ventral head region, deriving from the formation of extra eye fields on the eye disc (Jang et al. 2003). Forced toe expression in an eyg mutant background is also not able to rescue this phenotype. This suggests that these genes do not act redundantly in eye development and are differentially regulated, since eyg mutants do not affect toe transcription levels.
eyg is involved in a pathway independent of ey, since both genes are unable to launch each other’s expression. Moreover, they are also functionally independent, since they do not require each other to induce ectopic eyes (Jang et al. 2003). This functional independency resembles that of optix (Seimiya and Gehring 2000). Nevertheless, eyg and ey can functionally substitute for each other, by partially rescuing each other’s loss of function phenotype. Co-expression of these genes has synergistic effects in inducing ectopic eyes, and it has been postulated that they even form a heterodimer to commonly regulate gene expression (Jang et al. 2003) .
Hth encodes for the Drosophila homolog of the mouse Meis1 proto-oncogene containing a TALE class homeodomain. Hth provides an important function during embryonic patterning as a cofactor for the homeotic genes (Rieckhof et al. 1997; Pai et al. 1998).
hth is expressed in several imaginal discs including the eye-antennal disc. In the second larval stage, hth is ubiquitously expressed in both the eye and antennal part (Jang et al. 2003; Pichaud and Casares 2000). Later, the expression is restricted to the anterior parts of the disc including the ocellar region and head capsule. Weak expression can be detected in the posterior boundaries of the disc and in already differentiated pigment cells (Pai et al. 1998; Pichaud and Casares 2000).
hth mutants show ectopic eye formation in the ventral side of the head, suggesting that hth is normally needed to suppress eye formation (Pai et al. 1998; Pichaud and Casares 2000). The ectopic eye tissue induced in hth mutant clones have nearly normal pigment cells, but the ommatidia are abnormally arranged, showing an orthogonal shape instead of a hexagonal one (Pichaud and Casares 2000). However, in the dorsal region of the disc hth is not able to do the same. Forced expression of hth within the eye disc causes suppression of eye formation. The function of hth is dependent on its interactions with several RDN genes that will be described later in this chapter.
nmo encodes a proline-directed serine/threonine kinase and is the founding member of the Nemo-like kinase (NLK) family of the MAPK superfamily. In eye development it is required for positive regulation of its downstream target eya in the RDN cascade.
nmo shows a dynamic expression pattern in the eye disc. At second instar stage, nmo is ubiquitously expressed showing co-expression with ey in the whole disc, and with eya in the posterior part of the disc. Later in the third instar stage the expression becomes restricted to the anterior edge of the MF , showing co-expression with eya. nmo can also be detected in ocellar precursor cells at the anterior-dorsal side of the disc. At the posterior boundary of the disc nmo is expressed together with hth (Braid and Verheyen 2008).
nmo mutants have slim eyes and show defects in the patterning of the ommatidia (Choi and Benzer 1994). Strong forced expression of nmo leads to expansion of the dorsal eye, to the induction of dac and eya expression and to hth repression in the antennal discs, leading to ectopic PR formation. Ectopic PRs were not detected in other imaginal discs (Braid and Verheyen 2008). However, nmo acts synergistically with ey, eya, so, and dac to promote eye specification, and enhances their ability to form ectopic eyes (Braid and Verheyen 2008). nmo does not seem to be a direct regulator of these genes but rather interacts with them at the protein level.
distal antenna and distal antenna related
Dan and Danr are transcription factors containing a DNA binding motif called Pipsqueak (Psq) that is found in proteins involved in chromatin modification (Baonza and Freeman 2002). Misexpression of dan and danr with an antenna-specific driver Distal-less(Dll)-Gal4 leads to ectopic eye formation in antennal precursors (Curtiss et al. 2007). During the eye specification, dan and danr are particularly involved in the onset of differentiation, positively regulating the expression of one of the first differentiation genes atonal (ato). In addition, they interact with EGFR signaling to ensure correct ommatidial patterning (Curtiss et al. 2007).
dan and danr are required already during the early embryogenesis for the correct development of the nervous system. During larval stages their expression is restricted to the eye-antennal disc (Suzanne 2004). The expression in the early third instar eye disc starts in cells surrounding the MF. As the furrow starts moving, both factors are expressed anteriorly to it and to a weaker extent in the differentiated cells posterior to the furrow, whereby dan shows a stronger expression than danr (Curtiss et al. 2007). The expression of dan and danr overlaps with ey and shows similarities to eya, so, and dac expression (Curtiss et al. 2007). As both genes are later expressed also in the antennal disc, the timing of expression is critical in determining when these genes promote eye formation and when antennal formation (Curtiss and Mlodzik 2000; Suzanne 2004; Suzanne et al. 2003).
dan and danr loss-of-function mutants display small and rough eyes, depending on the severity of the mutation. The rough eye phenotype is a consequence of defects in ommatidial spacing and photoreceptor formation, which is probably associated with the lack of ato expression (Curtiss et al. 2007). Misexpression of dan and danr in the eye disc causes a similar phenotype to the loss-of-function mutants. This demonstrates the importance of a controlled regulation of transcription levels in a system that consists of many interconnected members like the one the RDN represents (Curtiss et al. 2007).
dan and danr are able to induce ectopic eyes in the antennal disc (Curtiss et al. 2007). In addition, as they are also expressed in the antenna, ectopic expression in the leg discs can induce leg-to-antenna transformation and interestingly, loss-of-function mutations cause an antenna-to-leg transformation (Emerald et al. 2003; Suzanne et al. 2003).
Dan and Danr cross-regulate each other; Dan is required for danr expression on the anterior side of the MF whereas Danr represses dan. Both genes are able to induce or maintain ey expression in the ectopic eye tissue (Curtiss et al. 2007). The discrete expression patterns and differences in eye phenotypes induced by dan and danr mutations propose that these two genes do not act totally in parallel during the eye formation, and that they might have even antagonistic roles (Curtiss et al. 2007). These two genes are acting downstream of most of the RDN genes, since so, eya, and dac are required for their expression (Curtiss et al. 2007) .
Teamwork: Genetic Interactions Between the Retinal Determination Network Genes
The eye of Drosophila represents a masterpiece when it comes to finding an example of a regularly patterned organ. To generate such a structure, a robust network of genes has to work behind it. The RDN offers an interesting and challenging model to study the early steps of tissue specification . The RDN consists not only of a linear, hierarchical cascade of transcription activation, but also it is imprinted by feedback loops, protein complexes, and dynamic gene expression(FIg. 2). Correct protein levels seem to be essential, especially in the formation of multiple protein complexes between the members of the RDN. In this part we will discuss the interactions of the RDN genes with each other and their putative target genes outside of the RDN.
toy and ey Initiate the Retinal Determination Network Cascade
toy is the first RD gene to be expressed in the eye precursors. Toy binds to the ey enhancer region and activates its expression. Forced expression of ey does not cause ectopic toy expression, placing toy upstream of ey in the genetic cascade (Czerny et al. 1999). ey rather than toy is more essential for the initiation of the eye development in general, since toy is not able to produce ectopic eyes in an ey mutant background (Czerny et al. 1999).
ey has been shown to be able to initiate so and eya expression in the wing disc when expressed ectopically. These genes contain Ey-specific binding sites on their enhancers, suggesting that Ey is a direct regulator of their expression. However, toy is able to initiate so expression even in mutants where the Ey-specific enhancer region is deleted (so 1 ), indicating that Ey and Toy can regulate so distinctively. Ey and Toy both bind the eye-specific so enhancer (so 10 ) with their paired domains, albeit at different sites (Niimi et al. 1999; Punzo et al. 2002). Surprisingly, in the few regions of antenna where forced so can induce ectopic eye formation, ectopically expressed ey and toy are not able to do the same. Therefore, there must be additional factors that mediate the interaction between ey/toy and so at least in the formation of ectopic eye tissue (Weasner et al. 2007).
Ey is required for normal eya activity, since eye discs mutant for ey do not express eya. Moreover, ectopic expression of ey induces eya expression (Bonini et al. 1997; Halder et al. 1998) and eya is able to induce ectopic eyes only in an intact ey background (Bonini et al. 1997). It is assumed that Ey binds to regulatory regions in the eya gene that have been shown to be responsible for the eye-specific eya expression (Bui et al. 2000; Zimmerman et al. 2000). However, although it was demonstrated with an electrophoretic mobility shift assay (EMSA) that eya is a direct target of Ey, no ey binding site was located in the enhancer regions required for eya expression in the eye disc. Moreover, this ey binding locus of eya does not drive eya expression in a reporter assay. Nonetheless, it cannot be excluded that other regulatory regions left out from this assay are also needed for the eye-specific expression of eya (Ostrin et al. 2006). This is supported by the observation, that the eya enhancer is expressed also in ey, so, and dac mutant backgrounds in a pattern comparable to wild type (Bui 2000).
Ey can also bind directly to the eye-specific enhancer of optix and forced expression of ey induces ectopic expression of optix (Ostrin et al. 2006). However, in hypomorphic ey 2 mutants optix expression is not affected in the eye disc and ectopic eyes are produced regardless of ey expression (Seimiya and Gehring 2000). Thus, it is not clear what the function of this regulatory site in the enhancer of optix is.
dac was shown to be downstream of ey, since ey expression is not lost in a dac mutant background, but to a limited level Dac is also able to induce ey expression in the antennal disc, suggesting an existence of a feedback loop (Chen et al. 1997).
So and Eya Regulate ey Expression in a Positive Feedback Loop
Since eya and so require Ey for the induction of ectopic eyes in tissues where ey usually is not present (Pignoni et al. 1997), Eya and So must be able to bind to an eye-specific enhancer on the ey gene. Indeed, the ey enhancer contains a So binding site (Niimi et al. 1999; Punzo et al. 2002). Physical interaction between Ey and So has also been demonstrated and these two genes do form a protein complex together (Pauli et al. 2005; Niimi et al. 1999; Zhang et al. 2006). This complex has been associated with ato regulation, linking eye specification directly with differentiation (Zhang et al. 2006).
eya acts downstream of ey, since ey expression is not affected in eya mutants and forced ey expression causes ectopic eya expression. Yet reciprocally, eya is needed for the formation of ectopic eyes induced by forced ey expression, since eye formation cannot be launched in tissues mutant for eya. Forced expression of eya alone is able to induce ectopic eyes only in an intact ey background (Bonini et al. 1997). Hence, Eya and perhaps also So are able to induce ey expression only as a complex and not individually.
So and Eya Form a Protein Complex Acting as a Transcriptional Regulator
So forms a transcription activating protein complex with Eya . Together they control multiple steps during the development of the eye disc towards differentiation by regulating cell proliferation, MF formation and propagation, and later during neuronal development (Niimi et al. 1999; Pignoni et al. 1997; Serikaku and O’Tousa 1994). Forced expression of this complex is able to produce ectopic eyes in the antennal, wing, and leg discs in an ey dependent manner (Pignoni et al. 1997). The potency of eya and so to induce ectopic eyes is much higher when these are simultaneously rather than individually misexpressed. As Eya lacks a DNA binding domain, it is likely acting as a transcriptional co-activator in this complex (Pignoni et al. 1997). Another possibility is that Eya is involved in regulating the phosphorylation states of So, hence influencing the activity of So (Weasner et al. 2007).
Forced expression of so induces the expression of eya and dac. It is not yet clear if the interaction is direct or not, but this finding has inspired the following proposal: Ey and Toy initiate the expression of so, which in turn activates eya expression. Further, So forms a complex with Eya regulating other downstream genes involved in the RDN (such as dac). It should be noted however, that So-induced eya expression is restricted to a certain subset of cells, suggesting the existence of additional factors in this pathway (Weasner et al. 2007). One of these factors could be Nmo (see the following section).
As previously mentioned, So alone physically binds the repressor protein Gro. The binding is, however, inhibited in the presence of Eya, although Eya and Gro are not competing for the same binding sites. This suggests that so has differential roles in (eye) development depending on the presence of its binding partner (Silver et al. 2003) .
eya and dac Have a Synergistic Effect on Eye Formation
eya was placed upstream of dac, since eya mutant eye discs show a reduced dac expression, and dac mutation does not affect eya expression. eya is not able to form ectopic eyes in the absence of dac and forced coexpression of eya and dac can induce ectopic eyes in a higher rate than when these genes are independently misexpressed (Chen et al. 1997; Tavsanli et al. 2004). eya is expressed in a nearly identical pattern with dac (and so), which led to the logical assumption that their gene products are molecularly interacting. Indeed, a physical interaction between Eya and the DD2 domain of Dac was demonstrated by in vitro biochemistry and yeast-one-hybrid experiments (Chen et al. 1997; Tavsanli et al. 2004). Unexpectedly, this physical interaction of Eya with the DD2 domain is not essential for the synergic effect observed by eya and dac co-expression, suggesting that Eya-Dac complex might not have any function in eye development (Chen et al. 1997; Kumar 2009). However, a possible trimeric complex formation between So, Eya, and Dac should not be excluded (Kumar 2009).
Nmo Promotes the Activity of the So and Eya Complex
Loss of one copy of eya in a nmo mutant background shows defects in the formation of the ventral eye and leads to an eye-to-head transformation. Overexpression of eya driver causes ectopic eye formation in the head; however, this phenotype is significantly reduced in nmo mutant background. The same is true for ey and dac. Overexpression of these genes cannot significantly reduce the small eye phenotype manifested by nmo mutants. Loss of nmo in general restricts the ability of these genes to induce ectopic eyes. In addition, loss of nmo dramatically reduces viability on dac mutants, causing their death as early larvae.
Reciprocally, overexpression of nmo together with eya enhances the formation of ectopic eyes. The synergistic relationship between eya and nmo is depended on the nmo kinase domain. Nmo-mediated phosphorylation of two MAPK sites on Eya promotes the activation of the Eya–So complex. Nmo is perhaps an intrinsic component of the Eya–So complex allowing fast and dynamic modulation of transcriptional output and could regulate the overall activity of the complex, but does this only in specific cellular contexts (Braid and Verheyen 2008; Morillo et al. 2012).
It is yet to be defined if Nmo first binds to a target DNA (trough other factors) and then recruits the complex or if the Eya–So complex recruits Nmo (Morillo et al. 2012). Co-expression of nmo and eya induce an increased expression of dac and lz. However, this increase is restricted into small dpp expression domains in the antennal and wing discs; hence, the role of the extracellular pathways should not be diminished. The increase in dac expression is also so-dependent, since in so mutants such an increase does not occur (Morillo et al. 2012).
tsh, tio, and hth Act Together to Suppress Eye Formation and Promote Proliferation
Tsh induces hth expression together with wingless (wg) signaling, and Tsh is required for the maintenance of Hth. Hth is a known repressor of eye development, thus acting as a mediator that causes eye suppression launched by tsh. Indeed, hth expression correlates with the severity of the split eye phenotype observed in some cases of tsh overexpression (Singh et al. 2002). hth and wg together suppress ventral eye formation, but do not affect the dorsal half of the eye (Pichaud and Casares 2000; Singh et al. 2002). Also Tio is able to maintain Hth protein levels after hth transcription ceases (Bessa et al. 2009).
Hth acts together with Extradenticle (Exd) in the same pathway by enabling nuclear localization of Exd (Pai et al. 1998). It was postulated that either Hth suppresses Dpp signaling or activates wg to suppress the initiation of the MF, or it interacts with a nuclear protein of the RDN (Pai et al. 1998). wg expression is lost in hth mutants, but only in the ventral head regions (Pichaud and Casares 2000). hth expression mimics that of wg, and they indeed are involved in a positive regulatory feedback loop in the ventral head capsule (Pichaud and Casares 2000). hth suppresses MF movement downstream of dpp (Pichaud and Casares 2000).
A complex consisting of Ey, Hth, and Tsh functions from the early eye disc to third instar larval stage to promote cell proliferation in the domain anterior to pre-proneural (PPN, region anterior to the MF) and thus suppress differentiation. This inhibition might also be a result of suppression of eya and dac in this region by this complex. Later in the PPN, where Hth is no longer present, it is likely that Ey, possibly together with Tsh, promotes eye formation by initiating so expression (Bessa et al. 2002).
Co-operation Between nmo and ey
It has been speculated that nmo contributes to the eye development by modulating the gene activity of other RDN genes , thus emphasizing the importance of the levels of gene transcription in the development of the eye. For example, loss-of-function mutants of nmo can compensate the severity of small-eyed ey loss-of-function mutant phenotype (ey R ), leading to bigger eyes, yet it acts together with Ey to induce ectopic eye tissue. It has been postulated that Nmo co-operates with ey in a context-dependent manner. In the first instar eye disc, they both promote eye specification by activating downstream genes, whereas in the third instar disc they would have antagonizing effects (Braid and Verheyen 2008).
Dan and Danr Act Physically With Ey and Dac Regulating the First Steps of Differentiation
After their expression has been initiated, Dan and Danr interact physically and genetically with their activators Ey and Dac. Dan and danr might function in a complex with Ey to regulate ato expression in a protein level-dependent manner, since misexpression of ey, dan, or danr lead to deformation of the eye (Curtiss et al. 2007). It is not clear what the role of the Dan-Danr-Dac complex is, but it is associated with the chromatin modifying function of Dan and Danr, making the chromosome around the eye-specific genes more accessible for transcription (Curtiss et al. 2007).
Identifying Targets of the Retinal Determination Network Genes
The compound eye of Drosophila is the result of the interactions of numerous genes, starting early in the embryo and lasting until the eclosion of the adult fly. The eye antennal imaginal disc expresses over 370 genes that are involved in the first steps of eye development (Michaut et al. 2003). What is the role of the RDN genes in the regulation of the rest? Are the RDN genes only controlling the early specification steps until the end of the larval stages or are they affecting the expression of the genes during the later steps as well, the ones that built the photoreceptors and confer them their function? Although a lot of progress has been made in the past years in understanding how the RDN functions, its targets remain still largely unknown. In the few studies conducted on RDN targets, microarray-based analysis of transcriptomes has been used as a tool. Two studies have searched for targets of ey by ectopically expressing ey in different imaginal discs and then comparing the gene expression patterns to wild-type discs (Michaut et al. 2003; Ostrin et al. 2006). The first study discovered 371 putative downstream targets of ey, without further analyzing them (Michaut et al. 2003). The second screen aimed to identify targets containing an Ey binding site, and found such sites, in so eya, optix, and shifted (shf), which is involved in the extracellular transport of the signaling molecule Hh. hh itself is important during the first steps of differentiation (Ostrin et al. 2006).
Nevertheless, targets whose expression depends on mutual expression of more than one RDN gene or on interaction with extracellular signaling cannot be detected with this approach. Recently, this was acknowledged in a study where forced expression of ey was combined either with Hh, Dpp, or Notch signaling (Nfonsam et al. 2012). Notably, a significant amount of putative targets would have remained undiscovered when the mutual overexpression would have not been applied. As it has become clear now that the RDN consists mainly of versatile feedback loops and interactions among its members, it is easy to appreciate the benefits this adapted microarray screen offers. Using this method, CG4721, a new target of ey with yet an unknown function was identified. This gene appears to play a role in the first steps of differentiation by regulating the expression of ato (Nfonsam et al. 2012).
A microarray-based screen was also conducted to search new targets for So–Eya complex. The screen revealed one new target, a cell cycle regulator String (Stg). stg was also upregulated in the first two ey target screens mentioned above (Michaut et al. 2003; Ostrin et al. 2006), probably because of Ey-mediated expression activation of eya and so and, thus indirectly, stg. Stg is a positive cell-cycle regulator (Jemc and Rebay 2007).
Similar screens would be necessary to reveal new targets of the other RDN members as well. Alternatively, sequence analysis has been relatively popular to compare known binding sites with putative target genes.
Eye Field Determination and the Primordial Cells of the Eye Imaginal Disc
While the adult imago emerges after metamorphosis of the larva, the primordia of adult structures are already set aside during embryogenesis. These primordial cells of the presumptive adult tissue are organized in imaginal discs. Imaginal disc cells are dedicated to form exclusively adult-specific structures, including the adult feeding organs, eyes, antennae, legs, halteres, wings, internal and external genitalia, as well as the epidermis. They develop stereotypically at defined positions in the larval body and can be identified by location, size, and developmental pattern in the larva. How and when are these cells in the embryo specified to form a particular structure?
Imaginal disc formation is initiated during cellular blastoderm stage (Simcox and Sang 1983). The discs originate from groups of founder cells at specific locations along the anterior–posterior body axes (Crick and Lawrence 1975). During early stages of embryogenesis, the eye field, giving rise to the eye-antennal disc, is formed as an elongated strip of cells in the developing dorsal pouch where the left and right primordium of the presumptive disc forms a V-shaped structure. The eye primordium can be identified by the expression of a Zinc-finger transcription factor Escargot (esg), which is required to maintain diploidy in imaginal disc cells (Hartenstein and Jan 1992; Hayashi et al. 1993). The presumptive eye-antennal imaginal disc develops from antennal, intercalary, and gnathal segments of the early embryo and also includes cells of the nonsegmental acron. Fatemap studies and lineage tracing have revealed that the eye-antennal imaginal disc originates from about 5–20 cells located at the anterior dorsolateral part of the early embryo. These cells form the presumptive eye field, which contains not only the eye-antennal disc but also most parts of the larval and adult visual system, such as the larval eye and the primordium of the adult optic lobes (Jurgens and Hartenstein 1993).
The disc is composed of two epithelial layers: the main epithelium (ME), a columnar layer also known as disc proper, and the peripodial epithelium (PE), a squamous epithelial layer (Haynie and Bryant 1986). The main epithelium gives rise to the eye, whereas the peripodial epithelium develops into the surrounding head capsule (Bessa and Casares 2005). The development of the eye-antennal disc proper takes place during late embryogenesis at stage 17, when the dorsal pouch shortens and the cells of the presumptive eye-antennal disc are compressed into a small cluster of cells. The inner layer of the dorsal pouch forms the medial wall of the eye disc, whereas the outer layer forms the PE. The PE participates in the fusion of the two bilateral symmetrical discs during metamorphosis (Fristrom et al. 1977; Pastor-Pareja et al. 2004). In freshly hatched larvae, the eye-antennal disc primordia are positioned in the third thoracic segment as a paired structure.
Genetic studies of several transcription factors and signaling pathways have identified the developmental program which is required for the proper specification and determination of the eye-antennal disc. ey and toy are expressed in the presumptive eye-antennal disc primordia and are both necessary and sufficient to turn the cell fate towards eye imaginal disc development. During embryogenesis, ey is expressed in a large domain covering the eye field giving rise to the eye imaginal discs and optic lobe primordia and in some distinct region of the brain and in VNC (Halder et al. 1995; Quiring et al. 1994). During early embryogenesis toy is expressed in a similar pattern as ey, however, toy expression precedes the expression of ey (Czerny et al. 1999). During the course of development, the toy expression domain gives rise to the brain and, if not all, to the most parts of the visual system including the optic lobe and eye imaginal disc primordia (Green et al. 1993; Younossi-Hartenstein et al. 1993). After germband retraction toy expression is restricted to the head region. At embryonic stage 13, toy expression marks the optic lobe primordia in a broader region that includes the brain and the presumptive eye imaginal discs. In contrast to toy, ey transcripts were first detected at stage 10 during germband extension in every segment of the developing VNC. However, during late embryogenesis at stage 16, both genes are expressed in different subset of cells in developing central nervous system. ey and toy seemed to be coexpressed in the optic lobe and the eye primordia of the late embryo (Czerny et al. 1999).
toy activation occurs very early during cellular blastoderm by the combined action of maternal active genes and gap genes. Based on the genetic analysis, a model has been proposed to demonstrate the onset of toy activation during embryogenesis. According to this model, after fertilization, maternally contributed bicoid (bcd) mRNA is translated generating a gradient of Bcd protein at the A/P axis with its highest peak at the anterior pole. This gradient activates transcription of hunchback (hb), which marks anterior part of the embryo and also restricts premature toy activation. During mid-cellular blastoderm stage, at the anterior pole, toy expression is suppressed due to the activation of knirps (kni) by the combined action of Bcd and Dorsal (Dl), which is a maternally contributed transcription factor and expressed as a gradient at the D/V axis. At the late blastoderm stage anterior part of the embryo is divided and the cephalic region is formed. toy is expressed in the cephalic region of the embryo by the combined action of Bcd, Torso (a maternally expressed Tyrosine kinase receptor) and Dl (Fig. 3). Thus, the genetic analysis shows that toy is activated at the cellular blastoderm stage as a result of a combined action of maternally contributed morphogens and zygotically expressed transcription factors (Blanco and Gehring 2008).
The Development of the Eye-Antennal Disc During First and Second Larval Stage: Determination of the Eye Primordia and Segregation of Eye and Antenna Fates
Developmental Plasticity Within the Eye-antennal Disc Is Maintained Until Second Larval Stage
While the disc giving rise to antenna and eyes originate from a single cluster of cells during embryonic development, within the eye-antennal disc, developmental plasticity is maintained until second instar stage. Clones induced up to second larval stage in the eye antennal disc could be found in any of the different structures the disc develops into (Morata and Lawrence 1979). Thus, at the time of clonal induction the cells marked by mitotic recombination do not segregate to the particular domain giving rise to the antenna or the eye. This is unusual compared to other discs, where the different domains are defined as early as blastoderm stage (Lawrence and Morata 1977). However, the exact timing of eye and antenna fate segregation has been subject to debate and has been placed either during late or early second larval stage (Kumar and Moses 2001; Kenyon et al. 2003).
The Onset of Gene Expression Patterns Correlates With the Late Eye-antennal Fate Segregation
The fact that the eye-antenna disc originally develops as a uniform field and is only late specified into different eye and antennal parts is reflected in its gene expression patterns of identified cell fate determinants (Fig. 3). Before fate segregation occurs, the early eye-antennal disc expresses the transcription factors that are necessary for both eye and antennal development. Only later, those genes become expressed in their respective domains. ey and toy are uniformly expressed in the early eye-antennal primordia during embryogenesis (Kammermeier et al. 2001). During second larval stage, the time of fate segregation between eye and antenna part, ey and toy are only expressed in the posterior domain of the disc, which marks the future eye part (Kenyon et al. 2003; Kumar and Moses 2001). Likewise, hth is expressed in the first larval stage in the whole disc but its expression becomes withdrawn from the posterior part during second larval stage (Bessa et al. 2002). Interestingly, hth remains expressed in the region of the eye disc that will develop into the head cuticle (Pai et al. 1998). Therefore, hth provides a later function during development by acting in the formation of sub-compartments within the eye disc.
The homeobox gene cut, essential for antennal development, is known to be the first marker of the antennal part of the eye-antennal disc (Bodmer et al. 1987). In contrast to ey and toy, it is not expressed in the eye-antennal disc before fate segregation but starts to be expressed exclusively in the anterior antennal domain by mid-second larval stage (Kenyon et al. 2003). cut expression is followed by the expression of Distal-less (Dll) (Kenyon et al. 2003). Both genes mark the future antennal part and are required for antennal development (Bodmer et al. 1987; Dong et al. 2000).
The Delayed Co-expression of Early Retinal Genes Locks in Eye Fate
A second group of genes starts to be co-expressed in the posterior eye field at the time of eye-antennal segregation during second larval stage (Kumar and Moses 2001). These genes, often referred to as early retinal genes, include the nuclear factors Eya, So, and Dac who work in a tight network specifying the formation of the eye (Desplan 1997; Kenyon et al. 2003; Kumar and Moses 2001). eya, so, and dac have been suggested to finally restrict the eye primordia’s competence for retinal differentiation upon the interaction with extracellular signaling pathways (Kumar and Moses 2001; Baker and Firth 2011).
Even though the eye selector genes ey and toy are expressed in the early disc, regional eye identity is only established during second larval stage with the co-expression of eya, so, and dac. How is the delay of regional eye fate achieved and how is the late onset of co-expression of the early retinal genes eya and so and dac established?
The Role of Extracellular Signalling Pathways in Separating Eye and Antennal Domains
Extracellular signaling pathways are repeatedly used in a spatial and temporal manner during Drosophila development (reviewed in Pires-daSilva and Sommer 2003). They include factors that define boundaries and axes or control cell proliferation. It is known that extracellular signals induce different developmental outcomes depending on the type of cells they act on (Baker and Firth 2011). Somehow these cells must combine the information that they receive and the developmental program that they are undergoing at a specific point in time. The Drosophila eye antennal disc serves as a model system to understand how the same extracellular signaling pathways acting in all imaginal discs results in different developmental outcomes. In case of the eye, how do the members of the RDN , specifying the early eye field, interact with these cell–cell signals to regulate Drosophila eye development?
While the distinct roles of signaling pathways in retinal differentiation and photoreceptor specification during third instar stage have been and still are extensively studied, their function and mechanism of action in the specification of the eye primordia during first and second larval stage has been less well characterized (Dominguez and Casares 2005).
Ectopically expressing genes of the RDN either alone or in combination can induce the formation of ectopic eyes. However, this ability is spatially restricted, which strongly indicates that additional factors are needed for proper eye formation (Kango-Singh et al. 2003). Most of the ectopic eyes are formed by overexpressing ey and are found in domains expressing dpp (Salzer and Kumar 2010). In addition, ectopically expressing ey with dpp and or hh (a potential activator of dpp) increases the range of cell populations that can transform into ectopic eyes (Kango-Singh et al. 2003).
dpp is a Drosophila homolog of the transforming growth factor β family (TGF β), which encode for secreted molecules acting in wide range of developmental processes (reviewed in Massague and Wotton 2000). There is evidence suggesting that dpp is required for the initiation of the early retinal genes during second larval stage. The expression of the early retinal genes eya, so, and dac is greatly reduced in second and third larval instar of dpp mutant eye discs, as has been shown by measuring their respective mRNA levels (Chen et al. 1999). The same is true for homozygous clone mutant for Mad, an effector of Dpp signaling pathway (Curtiss and Mlodzik 2000). Even though Dpp is necessary for activating so, eya, and dac expression, it is not required to maintain their expression (Curtiss and Mlodzik 2000). Originally it has been shown that ectopic eye formation by ey is observed in dpp expression domains (Chen et al. 1999). However, it was found recently that ectopic eyes can also form in cells outside of the dpp expression domain (Salzer and Kumar 2010). Even though this finding does not exclude the fact that dpp is required for normal eye development, it appears that the ability to form ectopic eyes does not require Dpp signaling.
The signaling molecule Wg is known to antagonize Dpp and to act as a suppressor of eye development (Hazelett et al. 1998). Ectopic expression of wg results in the loss of early retinal gene expression (Baonza and Freeman 2002). Moreover, loss of Wg signaling by blocking its receptors, results in ectopic expression of eya, so, and dac (Baonza and Freeman 2002). In line with the antagonistic actions of dpp and wg in controlling the expression of the early retinal genes, both genes are expressed in opposite sides of the eye-antennal disc in the early second instar larvae (Cho et al. 2000), wg along the anterior dorsal end and dpp along the posterior dorsal end (which will give rise to the future eye disc).
The Notch signaling pathway provides another essential function for proper eye development (Kenyon et al. 2003). The Notch receptor is activated by the ligands Delta and Serrate in the dorsoventral part of the eye disc, also referred to as the signaling centre of eye disc growth (Cho and Choi 1998). Notch activates the expression of eyg (Dominguez et al. 2004) which in turn activates the expression of unpaired (upd), a ligand of the Jak/Stat signaling pathway. As Upd is secreted, it may act over long distances to promote growth in the entire eye disc (Chao et al. 2004).
How do Notch, dpp, and wg signaling contribute to the late onset of early retinal genes and thus the final segregation of eye and antenna fates? It has been suggested that Notch signaling, contrary to previous findings, is not involved in directing the formation of eye and antenna fields by acting genetically upstream of ey (Kumar and Moses 2001). Notch signaling rather indirectly contributes to the regional specification of eye and antenna through its control of cell proliferation and disc size (Kenyon et al. 2003 Fig. 4). According to this model, cellular proliferation and thus, the increase of disc size initiated by Notch causes the opposing dpp and wg expression domains to be set apart. Cells that thereby no longer receive Wg signaling but Dpp in the posterior domain initiate the expression of eya. Confirming this model, reducing Wg signaling in small eye discs (initiated through antagonizing Notch) is sufficient to restore eya expression (Kenyon et al. 2003). However, this model does not explain the finding that eya and so are still expressed in eyg mutant eye discs that are severely reduced in size due to the prevention of eye growth (Dominguez and Casares 2005; Dominguez et al. 2004). Moreover, Eyg is negatively regulating wg expression, hence supporting eya and so expression, so there must be additional factors that are involved in the pathway suppressing wg.
Interestingly, similarly to the model described above, the proximal-distal segregation during vertebrate limb development has been proposed to be induced by cell proliferation (Tabin and Wolpert 2007). Through the proliferation of the limb morphogenetic field, the presumptive distal domains are moved away out of the range of the proximal signal retinoic acid. Thereby the future distal cells are now able to respond to the distal signal fibroblast growth factor. Taken together, these models shed a new light on how the morphogenetic field size can indirectly control the establishment of separate domains within an originally uniform field. In fact, it has been suggested that more complex body parts can only be formed during development by allowing proliferation and specification to interconnect (Amore and Casares 2010). Proliferation would enable cells to exit the regulatory state imposed by the original field and change their developmental fate. Therefore, studying the early developmental events in the formation of the Drosophila eye helps to understand more general schemes in developmental biology that appear to be common to many species.
Yet, it still remains to be investigated how ey and toy expression become restricted to the posterior part of the eye antennal disc as this occurs prior to the onset of dpp expression (Doroquez and Rebay 2006). The EGFR pathway has been suggested to suppress ey in the future antennal domain (Kumar and Moses 2001), which could explain the posterior restriction of ey and toy. In addition, EGFR has been shown to be required for the regulation of eya (Salzer et al. 2010). Thus, EGFR signaling has multiple roles during development depending on temporal and spatial clues. This makes the pathway difficult to investigate in early development by the simple study of loss of function phenotypes (Shilo 2003 2003).
Similarly, in segregating eye and antennal fates, the eye only develops from one of the two epithelia in the disc, the main epithelium (ME), even though, ey and toy are expressed also in the second layer, in the peripodial epithelium (PE). Tsh is a candidate transcription factor to restrict eye development to the ME as it is expressed only in this epithelium (Bessa and Casares 2005). When expressed ectopically in the PE, it initiates the expression of eya and dac, but is no longer able to do so when Dpp signaling is blocked. It appears that tsh enables the ME to mature into an eye by allowing it to respond to Dpp signaling (Bessa and Casares 2005).
Eye and Antenna Fates are Maintained by Mutual Repression
Once different eye and antennal primordial have been established, their fates have to be maintained. A way to achieve stable fate maintenance is to antagonize each other, for instance by the repression of transcription factors conferring the opposing fate. Indeed, it has been shown that the division of antennal and eye fate is preserved by the reciprocal repression between the eye and antenna determination genes (Wang and Sun 2012). cut and Hth repress ey transcription in the antennal domain by directly binding to its promoter. Similarly, So represses cut and hth in the eye domain of the disc.
While Ey and Toy are the earliest expressed transcription factors specifying the future eye disc during embryogenesis, the eye-antennal disc remains a uniformly specified field until second larval stage. During this stage the eye primordium becomes determined posteriorly by the expression of eya, so, and dac. Cell proliferation participates in the formation of an eye primordium that is competent to undergo retinal differentiation to form a fully differentiated compound eye.
The Role of the Retinal Determination Network in the Development of the Extraretinal Photoreceptors of Drosophila
The Retinal Determination Network in Larval Eye Formation
The Drosophila larvae sense light through the Bolwig’s organ, which is composed of 12 photoreceptor cells (Sprecher et al. 2007) . The developmental mechanisms controlling the larval and adult visual systems have some overlapping features. They both originate from the same ectodermal invagination during embryogenesis (Green et al. 1993) and they have the same pattern of photoreceptor axon projection through the optic stalk in a stereotypical pathway (Zipursky et al. 1984; Schmucker et al. 1997).
When comparing the genes involved in adult and larval visual system development, we find a number of RDN genes “shared” by these systems. The development of the larval eye begins with the invagination of optic lobe placode during stage 12 embryo (Green et al. 1993). toy is expressed in the entire developing eye fields including the presumptive larval eye and adult eye primordia during embryogenesis (Czerny et al. 1999), but toy is not required for the formation of the larval eye (Suzuki and Saigo 2000). Conversely, so has been shown to be required for both larval and adult visual system development (Serikaku and O’Tousa 1994). so is expressed in the larval eye precursors at stage 10 during embryogenesis and mutations in so lead to the absence of the larval eye (Cheyette et al. 1994). eya has been shown to be co-expressed with so at stage 10, and similar to so, no Bolwig’s organ is formed in eya mutants. Eya and So act together in a complex that activates ato expression in the larval eye precursors. ato is a proneural gene that is required for the development of the larval eye (Suzuki and Saigo 2000).
In summary, some of the retinal determining genes act not only during adult visual system development, but also seem to be important for the development of non-retinal visual organs like the larval eye.
The Retinal Determination Network in the Development of Ocelli
In addition to the compound eyes, the adult Drosophila has three simple light sensing organs called ocelli located on the dorsal head. These extra eyes are able to sense ultraviolet (UV) and blue light and serve as navigational help for the fly mainly during flying (Pollock and Benzer 1988; Yoon et al. 1996). The development of these visual organs begins in the third instar larvae from a dorsal anterior margin of the eye imaginal disc (Garcia-Alonso et al. 1996; Royet and Finkelstein 1996). This region not only gives rise to the ocelli, but to the whole vertex region including mechanosensory chetae and bristles (reviewed in Friedrich 2006).
In the developing eye imaginal disc, one gene, which is not part of the RDN , is exclusively expressed in the prospective ocelli forming region and is essential for the formation of the ocelli. The gene is called orthodenticle (otd) and belongs to the conserved otd/Otx gene family that is involved in the head formation both in insects and in vertebrates (Simeone et al. 1993). Otd is a transcription factor containing a homeodomain of the paired class (Finkelstein 1990). Viable otd mutants are ocelliless (Finkelstein et al. 1990). otd expression in the eye disc begins in the second instar larva and is restricted to the ocellar primordial cells by the third instar stage. This restriction is initiated by Wg and Hh signaling (Blanco et al. 2009). Subsequently, otd maintains its own expression via an autoregulatory feedback loop (Blanco et al. 2009).
The development of the ocelli is not as well understood as the development of the compound eye. Nevertheless, many of the RDN genes essential for the eye formation are also required in the specification of the ocelli. toy is one of the first genes expressed in the ocellar primordial cells (Brockmann et al. 2011) and toy mutants that are able to develop head and eyes are always lacking the ocelli (Jacobsson et al. 2009). ey transcripts are not present in this region (Brockmann et al. 2011). The regulation of the transcription initiation of toy in the embryonic eye primordium is still largely unknown, but otd seems to be involved in this process. Null mutants of otd lack toy expression in the eye precursor cells of the embryo, but the formation of the primordium itself is not affected, as one of the early determination genes, eyg, is expressed normally. otd expression is not abolished in toy mutants (Blanco et al. 2010).
otd and toy are not the only essential genes for the formation of the ocelli. Two other RD genes, so and eya, are also expressed in the precursors of the ocelli and mutations in these genes also lead to an ocelliless phenotype (Cheyette et al. 1994; Zimmerman et al. 2000). The expression of these genes is initiated during the third instar larval stage. The regulation of these genes in the ocellar region differs from their regulation during the retinal development. Toy was initially thought to be responsible only for so activation because of the presence of Toy binding sites in the so enhancer (Punzo et al. 2002). However, So is not totally lost in a toy mutant background. In addition, forced toy expression activates not only so but also eya expression. Toy is not able to initiate so expression in an eya mutant background, suggesting that eya is involved in so activation, as is the case in the compound eye formation (Blanco et al. 2010).
Although Wg is initially required for Otd expression,Otd itself represses wg. hh expression on the contrary is positively regulated by Otd, enabling eya activation. eya expression is then initiated by the activator form of Cubitus interruptus (Ci155), which is a transcription factor activated by the hh signaling pathway. Interestingly, Ci is not needed for the expression of eya in the compound eye (Blanco et al. 2009).
The current model suggests that Toy and Otd are involved in the initiation of eya expression and then together with Eya, perhaps by forming a protein complex, initiate the transcription of so (Blanco et al. 2010). Once initiated, so can maintain its own expression via an eya dependent autoregulatory loop (Pauli et al. 2005). Also, eya expression maintenance is so dependent (Brockmann et al. 2011).
It is intriguing how the RDN genes can act almost in parallel in two different genetic networks, having distinct responses depending on the genetic environment and cause different structural outcomes (retina and ocelli). This again enhances the view that the regulation of the RDN is highly dynamic and adaptable to the changing needs of the visual system of the fly during the course of evolution.
The formation of the Drosophila compound eye is a highly dynamic and stereotypically orchestrated developmental process. Using Drosophila as a model system, we can not only learn more about the similarities or differences in eye development between vertebrates and invertebrates, but the Drosophila eye serves as model to discover new principles of early developmental biology. Yet, many open questions considering the early eye development remain and certainly not all possibilities to study it have been explored. Although many signaling pathways and cell fate determinants acting during eye development have been identified and characterized, it is still surprising that a network of evolutionarily conserved transcription factors act as initial step to make an eye. While the genetic and molecular interactions between the individual players of the RDN are being studied in detail, the basic question of how the connections between the RDN members are orchestrated remains still largely unanswered. How are the members of this network able to induce ectopic eyes? This feature cannot be simply explained by the fact that the RDN genes induce the expression of genes involved in differentiation. If it was so, then why are the genes downstream of the RDN genes not able to do this? Despite the intensive research conducted in this field during the past two decades, we are still learning the rules of this game. The coming years will reveal how long it takes us to learn to play.
- de Zulueta P, Alexandre E, Jacq B, Kerridge S (1994) Homeotic complex and teashirt genes co-operate to establish trunk segmental identities in Drosophila. Development 120(8):2287–2296Google Scholar
- Friedrich M (2006) Ancient mechanisms of visual sense organ development based on comparison of the gene networks controlling larval eye, ocellus, and compound eye specification in Drosophila. Arthropod Struc Dev 35(4):357–378. doi:10.1016/j.asd.2006.08.010Google Scholar
- Fristrom JW, Fristrom DK, Fekete E, Kuniyuki AH (1977) The mechanism of evagination of imaginal discs of Drosophila melanogaster. Am Zool 17(3):671–684. doi:10.1093/icb/17.3.671Google Scholar
- Hartenstein V, Jan Y (1992) Studying Drosophila embryogenesis with P-lacZ enhancer trap lines. Roux’s Arch Dev Biol 201(4):194–220. doi:10.1007/BF00188752Google Scholar
- Heitzler P, Coulson D, Saenz-Robles MT, Ashburner M, Roote J, Simpson P, Gubb D (1993) Genetic and cytogenetic analysis of the 43A-E region containing the segment polarity gene costa and the cellular polarity genes prickle and spiny-legs in Drosophila melanogaster. Genetics 135(1):105–115PubMedGoogle Scholar
- Hoge MA (1915) Another gene in the fourth chromosome of Drosophila. Am Nat 49:47–49Google Scholar
- Jacobsson L, Kronhamn J, Rasmuson-Lestander A (2009) The Drosophila Pax6 paralogs have different functions in head development but can partially substitute for each other. Mol Gen Genomics 282(3):217–231. doi:10.1007/s00438-009-0458-2Google Scholar
- Jurgens G, Hartenstein V (1993) The terminal regions of the body pattern. In: Bate M, Martinez AA (eds) The development of Dorsophilia melanogaster, Cold Spring Harbor Laboratory Press, Cold Spring Harbor, pp. 687–746Google Scholar
- Kawakami K, Ohto H, Ikeda K, Roeder RG (1996a) Structure, function and expression of a murine homeobox protein AREC3, a homologue of Drosophila sine oculis gene product, and implication in development. Nucleic Acids Res 24(2):303–310Google Scholar
- Kawakami K, Ohto H, Takizawa T, Saito T (1996b) Identification and expression of six family genes in mouse retina. FEBS Lett 393(2–3):259–263Google Scholar
- Kenyon KL, Yang-Zhou D, Cai CQ, Tran S, Clouser C, Decene G, Ranade S, Pignoni F (2005) Partner specificity is essential for proper function of the SIX-type homeodomain proteins Sine oculis and Optix during fly eye development. Dev Biol 286(1):158–168. doi:10.1016/j.ydbio.2005.07.017PubMedGoogle Scholar
- Oliver G, Mailhos A, Wehr R, Copeland NG, Jenkins NA, Gruss P (1995a) Six3, a murine homologue of the sine oculis gene, demarcates the most anterior border of the developing neural plate and is expressed during eye development. Development 121(12):4045–4055Google Scholar
- Oliver G, Wehr R, Jenkins NA, Copeland NG, Cheyette BN, Hartenstein V, Zipursky SL, Gruss P (1995b) Homeobox genes and connective tissue patterning. Development 121(3):693–705Google Scholar
- Wolff T, Ready DF (1993) Pattern formation in the Drosophila retina. In: Bate M, Martinez-Arias A (eds) The development of Drosophila melanogaster, vol 2. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, pp 1277–1325Google Scholar
- Yao JG, Weasner BM, Wang LH, Jang CC, Weasner B, Tang CY, Salzer CL, Chen CH, Hay B, Sun YH, Kumar JP (2008) Differential requirements for the Pax6(5a) genes eyegone and twin of eyegone during eye development in Drosophila. Dev Biol 315(2):535–551. doi:10.1016/j.ydbio.2007.12.037PubMedGoogle Scholar
- Younossi-Hartenstein A, Tepass U, Hartenstein V (1993) Embryonic origin of the imaginal discs of the head of Drosophila melanogaster. Roux’s Arch Dev Biol 203(1–2):60–73. doi:10.1007/BF00539891Google Scholar