BIOLOGICAL OVERVIEW

The possibility that mechanisms of retinal determination may be similar between vertebrates and Drosophila has been supported by the observations that Pax6/eyeless genes are necessary and sufficient for retinal development. These studies suggest that the function of other gene families, operating during early eye development, might also be conserved. One candidate is the retinal homeobox (Rx) family of transcription factors. Vertebrate Rx is expressed in the prospective eye and forebrain and is required for eye morphogenesis, retinal precursor appearance, and normal forebrain development, indicating that it is an essential regulator of early eye and brain formation. The hypothesis that Drosophila Retinal homeobox (Rx) is required for adult and larval eye development was tested. When a Rx null allele was isolated, the mutant compound eye and larval visual system were not detectably abnormal. However, Rx is required for development of a central brain structure, the ellipsoid body, suggesting that Rx function in the brain may be conserved. Finally, a novel anterior head phenotype has been characterized; Rx is required for clypeus development. Thus, the data suggest that Rx may be required for the regulation of genes involved in brain morphogenesis and clypeus precursor development. It is proposed that differences in insect and vertebrate eye development may be explained by changes in gene regulation and/or the tissue of origin for eye precursor cells (Davis, 2003).

The Rx family of transcription factors contains a paired-like homeodomain and has been identified in chordates and invertebrates. Vertebrate Rx is expressed in the anterior neural plate (ANP), which gives rise to the eye and forebrain. Expression in the presumptive eye resolves to the neuroretina, while expression in the prospective forebrain becomes restricted to the hypothalamus, pituitary, and pineal gland. Loss-of-function mutations in mouse and medaka Rx result in severe early eye phenotypes (anophthalmia), in addition to defects in forebrain development. Retinal phenotypes are also observed in chicken and Xenopus embryos expressing dominant-negative Rx constructs. In addition, Rx is required for expression of Pax6 in the optic vesicle, consistent with an early role in establishing retinal precursors. Finally, overexpression of Rx induces ectopic retinal pigmented epithelium and duplications of the retina and neural tube. These studies are consistent with a role for Rx in the specification, patterning, and/or proliferation of eye and brain precursors within the ANP (Davis, 2003 and references therein).

Drosophila Rx (Drx) and vertebrate Rx share greater than 95% amino acid identity within their predicted homeodomains, indicating a high level of structural conservation (Eggert, 1998, Mathers, 1997 and Ohuchi, 1999). Drosophila rx is expressed in the procephalon, a region that gives rise to eye imaginal primordia and brain hemispheres. This pattern partly resembles vertebrate Rx expression in the ANP, suggesting that Drosophila Rx function has been conserved between taxa (Eggert, 1998; Mathers, 1997). However, although Drosophila rx is expressed in the embryonic brain, it has not been detected in eye imaginal primordia of late embryos or larval imaginal discs, suggesting that rx may have no function during eye development or that it functions earlier during the establishment of the eye primordia (Davis, 2003).

P-element-mediated mutagenesis and a local hopping strategy were used to generate Rx mutations. The ammunition element fam^k07505 is located ~85 kb proximal to Rx. An initial screen yielded a viable insertion, act57B^P(F5), located 431 bp upstream of act57B exon1. Using act57B^P(F5) in a second screen, a chromosome was isolated with two viable insertions: Rx^P(3A2), located 2.5 kb upstream of Rx, and the original act57B^P(F5). To generate deletions of the Rx locus, an excision screen was performed, using the double-insertion chromosome and Delta2-3 transposase. Lines were screened for loss of Rx sequences by PCR (Davis, 2003).

Excision line Rx^ex8 was selected for further analysis. In Rx^ex8 fly cultures, the majority of homozygotes fail to eclose and die as pharate adults. In the rare mutants that do eclose, their wings fail to unfold, they become dehydrated, and do not survive beyond 48 h. Unfolding of Rx^ex8 mutant wings with KOH demonstrates that there are no gross structural defects. Complementation analysis using deficiencies in the region demonstrates that the Rx^ex8 culture phenotype maps near the Rx locus. To confirm that loss of Rx function is responsible for the culture phenotype, attempts were made to rescue Rx^ex8 mutants with a Rx minigene, BSKK. While the majority of Rx^ex8 mutants fail to eclose, Rx^ex8BSKK/Rx^ex8 animals can successfully eclose, have straight wings, survive, and reproduce (Davis, 2003).

The clypeus is a part of the cibarium, which acts as a pump to draw food under the labrum, between anterior and posterior cibarial plates and into the esophagus. Suction is created by separation of the plates, due to contraction of muscles attached to the anterior cibarial plate and clypeus. To determine whether other parts of the oral pump were missing, both an external and internal evaluation of the Rx^ex8 cibarium was performed. Scanning electron microscopy reveals that the labrum is present in the mutants. However, analysis of H and E-stained sections from Rx^ex8 mutants reveals an abnormal cibarium. In controls, the anterior and posterior plates run parallel to each other and flank cibarial musculature. These muscles are organized by attachment to an apodeme, which is part of the anterior plate, and the clypeus . In frontal sections of Rx^ex8 mutants, the cibarial plates are present, as well as the anterior apodeme. However, the musculature is disorganized and does not appear to be consistently attached to the apodeme. These data demonstrate that Rx is not required for formation of the labrum or cibarial plates (Davis, 2003).

Transplantation studies have demonstrated that the clypeus, labrum, and cibarial plates are derived from the clypeolabral imaginal disc. Since Rx is required for development of the clypeus and is expressed in the embryonic clypeolabral bud, whether Rx is expressed in the larval clypeolabral imaginal disc was determined. Three criteria were used to identify the disc: (1) the clypeolabral disc is within the cephalopharyngeal skeleton (CPS); (2) imaginal disc cells are evident based on their small size compared with larval cells, and (3) a lacZ enhancer trap within the headcase locus is active in the clypeolabral disc. Using Rx and ß-galactosidase antibodies, immunohistochemistry was performed on the clypeolabral disc from headcase-lacZ larvae. Comparison of the Rx and ß-galactosidease staining patterns reveals that Rx is expressed in a subset of clypeolabral disc cells. This result correlates with a requirement for Rx in clypeus, but not the labrum or cibarial plates development (Davis, 2003).

Rx expression in the embryonic, larval and adult brain suggests that Rx may be required for brain development (Eggert, 1998). H and E-stained sections of Rx^ex8 mutant adult brains were examined. At the level of the central complex, analysis reveals an abnormal ellipsoid body (EB) in the mutants. In wild-type brains, the H and E-stained EB neuropile appears as a midline ring, dorsal to the esophagus. In contrast, the EB in Rx^ex8 mutants is disrupted. To confirm and extend these findings, immunohistochemistry was performed on frozen sections using FasII antibodies. In controls, the FasII-stained EB appears as concentric rings flanked by the mushroom body peduncle. In contrast, the FasII-stained EB rings are missing and are replaced by an elongated neuropile. Analysis using Leonardo antibodies, which also stain the EB, confirms that the Rx^ex8 mutant EB is malformed. These findings indicate that, in Rx^ex8 mutants, the EB is abnormally structured (Davis, 2003).

The GAL4/UAS system has been used to characterize the EB. Based on studies using multiple GAL4 drivers, the EB consists of large and small field neural processes that form distinct rings. Four main types of large field neurons, R1, R2, R3, and R4, have been identified. To examine the Rx^ex8 EB phenotype in more detail, the P{GAL4^c232} line (c232) was used; this line expresses GAL4 in R3 and R4 neurons, which contribute to the inner and middle ring, and outer ring, respectively. To directly visualize cell bodies and projections, a P{UAS-mCD8::GFP} GAL4 reporter gene (UCG), which expresses membrane-bound GFP, was used. In controls, the c232 cell bodies are located near the anterior surface of the brain. These neurons send projections toward the midline, through the lateral triangle and arborize into two concentric EB rings. This pattern does not vary significantly between control males, females, pharate, newly eclosed, or older adults. In contrast, in Rx^ex8 homozygotes, the pattern of EB arborization is severely disrupted. Although the mutant c232 cells project axons, which form a partial inner and outer ring, the rings fail to assemble at the midline (Davis, 2003).

During analysis of FasII-stained brain sections, variation was observed in the EB phenotype. To confirm this observation, c232 neurons of 47 mutant female brains were examined and compared. The results demonstrate 3 categories of EB phenotypes. The most severe 'unfused' phenotype, where termini fail to intersect at the midline, is observed in over half of the mutants (57%). A less severe 'elongated' phenotype is seen in one-third of the mutants, where the EB processes cross the midline, but are abnormal in width (28%). Finally, the least severe 'ventral' phenotype, where the EB width is normal, but the ventral processes fail to close the rings, is also the least frequent (15%). These phenotypes are observed in both sexes, in pharate adults and rare escapers (Davis, 2003).

To determine the role of Rx in brain development, the ability of a Rx minigene BSKK to rescue the c232 EB phenotype was tested. The Rx minigene is functional, since it can rescue the Rx^ex8 eclosion and clypeus phenotypes. The BSKK minigene can, at low frequency, completely rescue the EB phenotype. In addition, there is a significant reduction in the percentage of animals with the most severe 'unfused' phenotype (from 57% to 16%), and an increase in the percentage of animals with less severe 'elongated' (28% to 47%) and 'ventral' defective (15% to 33%) phenotypes. These data indicate an essential role for Rx in correctly forming and positioning the c232 EB (Davis, 2003).

In conclusion, the ellipsoid body in Rx mutants is abnormally formed, suggesting a conserved role for Rx during brain development. In addition, Rx mutants lack an anterior head structure, the clypeus, which is a component of the feeding apparatus. In general, these phenotypes correlate with the embryonic and larval expression patterns of Rx: while not detectably expressed in direct eye precursor cells, Rx is expressed in the developing brain and a derivative of the clypeolabral bud (Eggert, 1998; Davis, 2003).

Vertebrate Rx genes are expressed in strikingly similar patterns and have similar functions. In early vertebrate embryos, Rx is expressed in the anterior neural plate, including the prospective eye and forebrain. As might be expected, loss-of-function mutations in mouse and medaka Rx result in anophthalmia and severe forebrain defects. Similarly, expression of dominant-negative Rx constructs in chicken and Xenopus embryos results in eye and brain phenotypes. In addition, Rx is required for Pax6 expression in optic vesicle progenitors. Moreover, overexpression studies demonstrate Rx can induce ectopic retinal development. These studies indicate a role for Rx in establishing the early eye field in such diverse species as fish, amphibians, birds, and mammals (Davis, 2003 and references therein).

Although Rx and vertebrate Rx share some similarities, these genes differ in their requirements during establishment of the eye. The Drosophila Rx homeodomain shares greater than 95% amino acid identity with vertebrate Rx homeodomains, indicating a high level of structural conservation. In addition, Rx expression in the embryonic procephalic region bears a resemblance to vertebrate Rx expression in the ANP. However, the visual system in Drosophila Rx mutants is normal and gross abnormalities were detected. Since Rx^ex8 is a null allele and no other Rx homologs are in the sequenced fly genome, there is no residual Rx function in the Rx^ex8 flies. Recently, planarian Rx homologs from Dugesia japonica and Girardia tigrina have been cloned, but they are not detectably expressed in the eye (Salo, 2002). These results suggest that the utilization of Rx function in the development of light-sensing organs of some invertebrates has not been required since the divergence of vertebrata from a common bilaterian ancestor. However, it is still possible that Rx plays some role in eye function or may be required for aspects of eye development that have not been detected by the methods that were used to examine the mutant phenotype (Davis, 2003).

Evidence has been forwarded that the topology map of the embryonic brain/eye anlage in Drosophila is similar to the fate map of the vertebrate ANP. This map is based on the effect of hedgehog and decapentaplegic mutations on patterning the expression of conserved genes implicated in brain and eye development, including orthodenticle, tailless, eyeless, eyes absent, and sine oculis. However, comparison of vertebrate Rx and Rx expression patterns indicates a level where the topology map is not conserved. While vertebrate Rx expression in the ANP encompasses forebrain and retinal anlage, Rx expression appears within the brain anlage but probably not in the region giving rise to the visual system. Moreover, the data confirm the absence of a requirement for Rx in the gross development of the adult and larval visual systems. Thus, the lists of genes required for establishment of the fly and vertebrate eye anlage are not identical, despite the otherwise striking parallels in embryonic head patterning (Davis, 2003 and references therein).

There are many possible explanations for a 'breakdown' within the conserved topology map, including differences related to the development of the eye between insects and vertebrates and/or changes in Rx regulation during evolution. The vertebrate retina is derived from the neuroepithelium, while the Drosophila compound eye develops from embryonic surface ectoderm, which is set aside as an imaginal disc and develops as retinal tissue during larval stages. In this sense, the insect eye shares more character with the vertebrate lens, which is also derived from the surface ectoderm adjacent to the neuralized epithelium. It is interesting to note that expression of vertebrate Rx genes has not been detected in the developing lens. One possibility is that the tissue-specific regulation of Rx has remained the same in insects and vertebrates, but that the pool of cells recruited to become the primary light-sensing organ differs between animals. Alternatively, some aspects of Rx regulation may have changed since the divergence of ancestral bilaterians or Rx may not have been recruited for function in the invertebrate visual system due to compensatory involvement of other genes (Davis, 2003).

Other aspects of Rx expression suggest conservation of Rx function in the brain. In particular, Rx expression in the brain anlage is in close proximity to, or contained within, orthodenticle and tailless expression domains, which is similar to the expression profiles of vertebrate Rx, Otx2, and Tlx in the prospective forebrain. This suggests that some modes of Rx/drx regulation by neuralizing factors may have originated in ancestral bilaterians and remain mostly unchanged (Davis, 2003 and references therein).

Analysis of the Rx^ex8 allele reveals ellipsoid body (EB) abnormalities within the mutant adult brain. The EB is a circular neuropile that is part of the central complex, which includes the mushroom body, fan-shaped body, noduli, and protocerebral bridge. The EB has been implicated in the control of specific locomotor skills, since mutations that affect EB structure are associated with changes in walking and flight activities. No EB embryonic precursor has been identified, but an EB-specific GAL4 line has been used to demonstrate that ring development begins around 32 h after puparation formation (APF) and is completed by midpupal stage (~48 h APF). GAL4 drivers have been used to analyze R2, R4m, and R4d EB neurons in ellipsoid body open (ebo), central body defect (cbd), and central complex deranged (ccd) mutants. ebo mutant EB neuropiles are fused at the midline, but lack variable aspects of the ventral ring. cbd mutant EBs frequently fail to join at the midline and display ventral defects. Finally, ccb mutants EB exhibit a range of phenotypes, from duplication at the midline to ventral defects (Davis, 2003).

R3 and R4d EB neuropiles were analyzed in Rx^ex8 mutants, using the c232 GAL4 driver, and phenotypes similar to ebo, cbd, and ccd: unfused, elongated, and incompletely fused EBs, were found. In the most severe phenotype, Rx^ex8 mutant EBs fail to fuse at the midline, which is similar to the cbd phenotype, and consistently lack a ventral aspect similar to the ebo and ccd mutant EBs. However, Rx^ex8 mutants also exhibit an 'elongated' EB, which has not been documented in ebo or cbd mutants. In each instance, the EB ring is incomplete along the ventral aspect, suggesting that ring completion is more sensitive to genetic perturbation than ring positioning. These EB phenotypes may be due to differences in the ability of growth cones to respond to midline cues. Alternatively, brains with mutant EBs may be missing pioneering tracts or supporting glia that are required for ring positioning. Finally, loss-of-function mutations in these genes may affect similar, but nonidentical sets of neurons that constitute the EB (Davis, 2003).

Rescue analysis indicates that both Rx and act57B transgenes reverse the Rx^ex8 EB phenotype in a similar manner. This picture contrasts with activities of these transgenes in other assays. While the Rx transgene rescues the Rx^ex8 culture and clypeus phenotypes, the act57B transgenes fail to do so. Conversely, the act57B transgene can rescue the prepupal lethality in Rx^ex8/E2 mutants, but the Rx transgene cannot rescue this phenotype. Taken together, these data indicate that Rx and act57B are both required for EB development. When the transgenes are combined in the same fly, their effect on EB rescue is additive but not synergistic, consistent with the possibility that the two genes act independently (Davis, 2003).

At this early stage of analysis, it is difficult to precisely define the mechanisms by which Rx and act57B function in EB development. However, some predictions can be made based on the expected properties of these proteins. act57B encodes an actin isoform expressed in embryos, larvae, and adults, and no mutations in this gene have previously been described. Since cell motility depends on actin polymerization dynamics, it is possible that the Rx^ex8 EB phenotype is partly due to abnormal act57B levels; this results in a cell population that fails to migrate/project axons correctly within the brain. Indeed, the actin binding protein Ciboulot is important in regulating cytoskeleton dynamics and is required for normal EB development. Of interest, the ciboulot mutant EB phenotype is similar to the ventrally defective Rx^ex8 EB phenotype, suggesting that regulation of act57B polymerization by Ciboulot may be required for axons closing the EB ring (Davis, 2003 and references therein).

In contrast, Rx encodes a putative transcription factor, suggesting a role regulating gene expression in the developing brain. It is possible that, in Rx^ex8 mutant brains, the absence of Rx results in changes in the specification, differentiation, and/or growth of cells that are required for normal EB development. Other transcription factors, such as AP-2, Dachshund, and Eyeless, are also required for normal EB development. Since Rx and act57B appear to act in different pathways, it is unlikely that Rx directly regulates act57B. The isolation of gene-specific mutations is required to characterize the individual roles of Rx and act57B in the formation of the EB. Once gene-specific mutations are generated, it will be possible to precisely assign other genes to either the Rx or act57B pathways (Davis, 2003).

Larval transplantation studies have demonstrated that the clypeolabral imaginal disc lies within the cephalopharyngeal skeleton (CPS) and gives rise to the adult clypeus, labrum, and cibarial plates, which form components of the oral pump (cibarium). Histological analysis using an imaginal disc marker, headcase-lacZ, has confirmed the presence of a single pair of discs inside the CPS. Beyond this, however, little information about the disc or its derivatives is available. No fate mapping studies or developmental characterization of the disc has been conducted, most likely a result of a lack of useful adult markers and the disc's intractable location. Finally, no mutations have previously been described that specifically affect the adult structures derived from the disc (Davis, 2003).

This work represents an initial step toward understanding this obscure aspect of fly development. In Rx^ex8 mutants, the clypeus is missing, while the labrum and cibarial plates are present. The findings of a requirement for Rx in clypeus development and Rx expression in a subset of the disc suggest that Rx is required for clypeus precursor cell development. There are two possible functions of clypeus cells. Since the clypeus is an exoskeleton element, they may secrete cuticle to form this structure. In addition, since the clypeus is an attachment site for the underlying cibarial musculature, the cells may act as 'tendon cells' by forming attachments with myotubes. The absence of an external clypeus and abnormal cibarial musculature in Rx^ex8 mutants suggests that Rx is required for both of these functions. Since the anatomy of the clypeus has not been characterized in detail, it is not clear whether these functions are mediated by a single cell type or multiple cell populations (Davis, 2003).

The absence of a clypeus in Rx^ex8 mutants likely results in an ineffective cibarium, which may explain three aspects of the culture phenotype: failure of pharate adults to eclose, and in rare escapers, folded-up wings and dehydration. The frontal ganglion innervates the cibarial muscles of manduca sexta and is required for the cibarial motor program, feeding, eclosion, and wing expansion. Surgical ablation of the frontal ganglion in pupae results in the majority of pharate moths failing to eclose, and among those that do eclose, their wings fail to expand. It is believed that the motor activity of the cibarium is required for swallowing fluids before eclosion and swallowing air afterward, which may be important in generating abdominal pressure necessary either for escape from the pupal case or forcing hemocoel into the wing veins, respectively. The Rx genomic transgene rescues both the clypeus and culture phenotype, suggesting that Rx function in the clypeus is sufficient to restore cibarial function and reverse the culture phenotype. However, it is possible that other Rx functions, not associated with the clypeus, are required for eclosion. In the future, the generation of a clypeus-specific GAL4 driver may be useful in testing this hypothesis. These studies have now generated reagents and a conceptual foundation to begin to explore the underlying mechanisms of Rx function in brain and clypeolabral development (Davis, 2003).

GENE STRUCTURE

Two alternatively spliced transcripts of Rx were detected. The combined cDNA sequence of the longer splice form DRx 1 has a total length of 3.2 kb. The ORF starts at position 269, terminates at position 2974, and encodes a protein of 902 amino acids with a predicted molecular weight of 95,757 and an isoelectric point (pI) of 6.81. The sequence preceding the ATG fits only poorly to the consensus translation initiation sequence. The termination codon TAG found at position 903 is followed by a putative polyadenylation signal (AATAAA) and a 26-nucleotide poly(A) tract. The transcription unit of the Rx gene consists of seven exons spanning a genomic region of at least 18 kb. The transcription initiation site remains to be determined. The homeodomain comprises two exons with an intron at position 44, a very common splice position for homeodomain proteins. The intron size between the two homeodomain exons is about 9 kb. Alternative splicing in the 3' part of the gene results in a putative protein form that is 130 amino acids shorter (Eggert, 1998).

cDNA clone length - 3242 bases

Bases in 5' UTR - 361

Exons - 6

Bases in 3' UTR - 258

PROTEIN STRUCTURE

Amino Acids - 902

Structural Domains

Highly conserved domains in the Drosophila Rx protein are the octapeptide, the identical homeodomain, the carboxyl-terminal OAR domain, and a newly identified Rx domain. In the carboxyl-terminal part of the deduced protein sequence, a paired-like homeodomain was identified that is 100% identical to the homeodomain of the Rx gene from Xenopus laevis, 98% identical to the Rx gene from mouse, and 97% identical to the Rx gene from zebrafish. The same homeodomain sequence was also reported as bk50 in a screen for homeodomain proteins binding to a common Engrailed binding site (Kalionis, 1993). In contrast to paired-type homeodomains found in Pax genes that share a characteristic serine residue at position 50, paired-like homeodomains have a glutamine at this position. Like the vertebrate genes, Rx has an octapeptide sequence in the amino-terminal part and an OAR domain at the carboxyl terminus. Additional sequence conservations are found at the amino and carboxyl termini of the homeodomain and in a region between the homeodomain and the OAR domain. This latter region has been designated as the Rx domain. However, the total Rx protein is more than twice as long as the corresponding proteins in vertebrates and very rich in alanine (8.4%), glycine (9.9%), serine (10.4%), threonine (4.9%), and proline (10.0%) (% values are molar). Glutamine (7.8%) is mainly present in the form of an M or opa repeat (Eggert, 1998).

date revised: 2 November 2003

Home page: The Interactive Fly © 1995, 1996 Thomas B. Brody, Ph.D.

The Interactive Fly resides on the
Society for Developmental Biology's Web server.