mirror: Biological Overview | Transcriptional Regulation | Targets of Activity | Developmental Biology | Effects of Mutation | References

Gene name - mirror

Synonyms -

Cytological map position - 69D1--69D6

Function - transcription factor

Keywords - eye, segment polarity

Symbol - mirr

FlyBase ID: FBgn0014343

Genetic map position - 3-

Classification - homeodomain - Pbx class

Cellular location - nuclear

NCBI links: Precomputed BLAST | Entrez Gene

The eye is composed of dorsal and ventral fields of photoreceptor clusters called ommatidia. Ommatidia in the dorsal half of the eye are the mirror image of those in the ventral half. mirror (mrr), a component of the iroquois locus that also includes araucan and caupolican, is expressed in the dorsal half of the eye imaginal disc; as such it is a primary determiner of the border between dorsal and ventral halves of the eye (see Specification of the eye disc primordium and establishment of dorsal/ventral asymmetry) and has a marked effect on eye symmetry (McNeill, 1997).

The boundary where dorsal and ventral fields meet is known as the equator. The equator bisects the eye from anterior to posterior, rarely deviating by more than one ommatidial width as it crosses the eye. Preclusters (the progenitors of ommatidia) rotate in opposite directions as they mature in dorsal and ventral regions. Asymmetries are incorporated into the preclusters as they rotate, resulting in ommatidia of opposing polarity and chirality (handedness) in the adult eye. Recent work has demonstrated that the degree of ommatidial rotation is controlled by several genes, including nemo and roulette, and has suggested that tissue polarity genes such as spinylegs, prickly-spinylegs, frizzled, and dishevelled are necessary for regulating DV polarity in the eye (McNeill, 1997 and references).

The equator might be the source of D/V patterning information. For example, frizzled is a transmembrane protein required to interpret and relay a signal emitted by the equator (Zheng, 1995). A number of genes appear to be expressed either solely at the equator or in a gradient from the equator. Four-jointed, either a cell surface or a secreted protein, is expressed in a graded fashion around the equator; mutants show a variable deformation of the eye surface. Morphological analysis of normal eyes indicates that ommatidial differentiation in each row proceeds from the equator onward. It seems that more mature ommatidia provide D/V polarity information to their less differentiated neighbors (McNeill, 1997 and references).

The iroquois locus was initially defined by iroquois mutation, isolated in a search for mutations that alter the pattern of macrochaetae (Leyns, 1995). Two transcriptions units, araucan and caupolican, were initially detected within the iroquois locus The proteins possess homeodomains similar to those of Drosophila Extradenticle and mammalian Pbx1 (Gómez-Skarmeta, 1996). Subsequently, the gene mirror was identified, closely linked to ara and caup. mrr was identified using a P element insertion element that carried two markers, white and lacZ, whose expression is restricted to cells in the dorsal half of the eye. The expression patterns of P element vectors is determined by the regulatory elements of the gene into which the vector inserts. Thus the expression pattern of white and lacZ potentially marks a promoter directing expression of a gene to the dorsal half of the eye. Mirror also proves to be a homeoprotein, and is the third PBX-class homeoprotein of the IROC complex (McNeill, 1997).

Since strong mirror mutants show an embryonic or early larval lethality, the function of mirror in the eye was examined by generating loss-of-function clones. Loss of mirror function has different effects in different parts of the eye. There are no effects in the ventral half, consistent with the observation that mirror expression is restricted to the dorsal half of the eye. Dorsal clones also have no significant differences in ommatidial polarity or chirality within the clone. These results suggest that mirror is not required for the implementation of dorsal identity. There is, however, a dramatic alteration in patterning at the border of certain clones. Ectopic equators form at the equatorial borders of dorsal anterior clones where mirror-minus tissue abuts mirror-expressing tissue. The wild-type ommatidia just outside the clone adopt ventral polarity and chirality. Ventral patterning extends for one to two ommatidial widths, then the tissue resumes normal dorsal patterning. The formation of ectopic equators where mirror-expressing and -nonexpressing cells meet suggests that the juxtaposition of mirror-expressing and -nonexpressing cells serves to define the normal equator (McNeill, 1997).

An examination of mirror-minus clones that cross the normal equator reveals that where the equator approaches a patch of mirror-minus tissue, the path of the equator is diverted to follow the new border of mirror expression. After taking several dorsal steps paralleling the border of mirror-expressing and -nonexpressing cells, the equator resumes its posterior to anterior path across the eye rather than continuing to follow the edge of the clone. This result suggests that the pattern of mirror expression is not the sole determinant of equator position and that there may be spatial restrictions on the ability of mirror to define the equator (McNeill, 1997).

The mechanism of D/V boundary determination in the wing may provide a model for understanding equator formation in the eye. Dorsal identity in the wing is controlled by the homeodomain protein Apterous, which is expressed solely by dorsal cells. The juxtaposition of Apterous-expressing and -nonexpressing cells defines the D/V border in part by directing dorsal expression of the secreted protein Fringe. Fringe then directs patterning at the D/V border in part by controlling the Notch pathway. It will be interesting to see if Apterous and mirror are found to control some of the same genes (McNeill, 1997).

Mirror also is involved in segmentation. The onset of mirror expression in each segment comes after the establishment of parasegmental signaling, but prior to and during the determination of the segmental border and the refinement of Engrailed and Wingless expression. As the border between Engrailed- and Wingless-expressing cells determines the parasegmental boundary, so might the juxtapositioning of mirror-expressing and -nonexpressing cells be important for defining the segmental border. Consistent with this hypothesis is the obervation that mirror mutant embryos have segmental patterning defects. mirror might therefore function both in the eye and during segmentation to define borders (McNeill, 1997).

Many genetic cascades trigger different responses and hence determine different cell fates at specific times and positions in development. At stage 10 of oogenesis, mirror is expressed in anterior-dorsal follicle cells, and this is dependent upon the Gurken signal from the oocyte. The fringe gene is expressed in a complementary pattern in posterior-ventral follicle cells at the same stage. Ectopic expression of mirror represses fringe expression, thus linking the epidermal growth factor receptor (Egfr) signaling pathway to the Fringe signaling pathway via Mirror. The Egfr pathway also triggers the cascade that leads to dorsal-ventral axis determination in the embryo. twist was used as an embryonic marker for ventral cells. Ectopic expression of mirror in the follicle cells during oogenesis ultimately represses twist expression in the embryo, and leads to phenotypes similar to those that occur due to the ectopic expression of the activated form of Egfr. Thus, mirror also controls the Toll signaling pathway, leading to Dorsal nuclear transport. In summary, the Mirror homeodomain protein provides a link that coordinates the Gurken/Egfr signaling pathway (initiated in the oocyte) with the Fringe/Notch/Delta pathway (in follicle cells). This coordination is required for epithelial morphogenesis, and for producing the signal in ventral follicle cells that determines the dorsal/ventral axis of the embryo (Zhao, 2000).

mirror is expressed in a dynamic pattern during oogenesis. This includes expression in the germarium and in centrally-located follicle cells at stage 6 and the anterior-dorsal and centripetal follicle cells at stage 10. This paper concentrates on the regulation and function of mirr expression at stage 10 of oogenesis. At this time, mirr is expressed in the cells in the follicular epithelium that receive the Grk signal. This signal is known to be required in these cells to establish the dorsal-ventral axis of the egg and embryo. In grk-mutant ovaries mirr expression disappears or is occasionally reduced to a small group of follicle cells at the dorsal midline. In fs(1)k10 mutant ovaries, where grk transcripts diffuse from anterior-dorsal towards anterior-ventral within the oocyte at stage 9, all anterior follicle cells receive the Grk signal and an expanded expression of mirr is observed. Thus mirr expression is downstream of the Grk signal, and those cells that have activated Egfr signaling at this stage of oogenesis appear to respond by activating this homeodomain transcription factor. It is still not clear whether the early expression of mirr in the germarium requires the Grk/Egfr signaling. Grk/Egfr signaling activity does not always activate the mirr gene, since mirr is not expressed in posterior follicle cells at stage 7 of oogenesis, when grk signaling occurs at the posterior of the egg chamber. Other factors must, therefore, determine the way in which follicle cells, in different positions and with different developmental histories, respond to the grk signal (Zhao, 2000).

fringe (fng), which encodes a glycosyltransferase-like secreted protein, is involved in different developmental processes, such as the development of the wing and the eye and oogenesis. Its expression at stage 10 of oogenesis is restricted to the ventral and posterior follicle cells, and it is not observed in those cells that express mirr. So the expression patterns of mirr and fng are complementary, and both these patterns are defined by the position of Egfr activation in response to the Grk signal (Zhao, 2000).

To test whether the complementary expression of mirr and fng depend upon Egfr signaling at stage 10, their expression patterns were examined in flies expressing activated or dominant negative forms of Egfr. When the activated form of Egfr (DERAF or lambdatop) is expressed in the anterior follicle cells, all those cells now express mirr, but not fng. When the dominant negative form of Egfr (DER DN) is expressed in those follicle cells surrounding the oocyte, but not the centripetal follicle cells, the anterior-dorsal follicle cells no longer express mirr. The expression domain of fng expands to include the anterior-dorsal cells in these mutants. Thus, the expression of both mirr and fng are either positively or negatively regulated by the activation of Egfr, and a complementary expression pattern is maintained in all these experiments. Experiments show that Grk/Egfr signaling is required to activate the expression of mirr in anterior-dorsal and centripetal follicle cells, which in turn represses the expression of fng in those cells. As a result, fng is only expressed in the posterior and ventral follicle cells, where it is required for the normal morphogenesis of the follicle cell layer (Zhao, 2000).

fng expression is restricted to specific regions of wing discs, eye discs and follicle cells of the egg chamber. The repression of fng expression in anterior-dorsal follicle cells in wild-type egg chambers by mirr possibly uses a similar mechanism to that observed in eye development, where it is required for the modulation of the dorsal-ventral boundary established by Notch activation. As a secreted protein, Fng modulates the binding of Notch to its ligands at the dorsal-ventral boundary. Ectopic expression of fng induces new dorsal-ventral boundaries in the wing disc and can reverse the planar polarity of photoreceptor clusters in the eye disc. Interfering with fng expression using antisense RNA experiments and mitotic clones, causes abnormalities in epithelial development in the egg chamber and defects in the positioning of the chorionic appendages. Vertebrate Fng homologs are similarly involved in mediating the signals between dorsal and ventral cells during limb development. These findings suggest that the boundary between fng-expressing and non-expressing cells is important in pattern formation, and the restricted expression pattern is regulated by mirr, at least in eye development and oogenesis in Drosophila. In this way, mirr is controlling epithelial morphogenesis via fringe and possibly other targets. As yet it is not known whether mirr directly represses fng transcription or whether there are other genes in the pathway between mirr and fng (Zhao, 2000).

fng affects eggshell patterning. The Gal4/UAS system was used to misexpress mirr, and this results in abnormalities of the chorion. When mirr is ectopically expressed at low levels in follicle cells surrounding the oocyte, there are enlargements of the dorsal appendages in the eggs, some of which (2%) become multiple pseudo-dorsal appendages. During oogenesis, when the ectopic expression of mirr is driven by the Gal4 line C710, 95% of eggs laid by these females have no chorion, suggesting that ectopic expression of mirr can repress the expression of the genes in the ventral cells needed for secretion of the chorion over the ventral regions of the egg. Thus mirror is needed for the correct dorso-ventral patterning of the eggshell which is secreted by the follicle cells. How much of this is mediated via fng, and how much through other target genes and pathways is not known (Zhao, 2000).

Eggshell polarity and embryonic polarity are closely integrated and grk signaling initiates both these processes. Some genes downstream of the grk signal affect patterning in both the eggshell and embryo, e.g. Egfr, while others just affect the eggshell, e.g. BR-C. Therefore, it was asked whether mirr also affects patterning in the embryo. Both patterning events can be monitored in individual eggs by observing the phenotype of the egg chorion, which is secreted by the follicle epithelial layer, and by the cuticle pattern of the resulting embryo. In those eggs with overgrowths of the dorsal appendage, induced by misexpression of mirr, the denticles of the resulting embryos are partially or completely lost, showing a dorsalized phenotype. This suggests mirr affects both patterning systems (Zhao, 2000).

To further investigate how the dorsal-ventral polarity of embryos is affected by ectopic expression of mirr, expression of the twist (twi) gene was used as a marker for ventral embryonic cells. After Egfr activation in anterior dorsal follicle cells, 11 dorsal-group maternal genes are later involved in the establishment of a gradient of nuclear Dorsal (DL) protein in the ventral cells of the embryo. This in turn activates the expression of twi in the whole ventral domain of the wild-type embryo. In those eggs laid by females with ectopic expression of mirr driven by the Gal4 line T155, the central expression domain of twi is reduced and in some embryos completely disappears. The terminal domains of twi at both ends of the embryos are rarely affected. Similar phenotypes are observed when DERAF is ectopically expressed, but the frequencies of abnormalities are very different. Of the eggs laid, 98% have multiple dorsal appendages, rather than the 2% observed following ectopic expression of mirr. Also, most eggs lose the main domain of twi expression following ectopic expression of DERAF. The terminal regions of twi expression still remain, using this Gal 4 driver for DERAF (Zhao, 2000).

Further evidence that mirr is responsible for the repression of ventral genes that direct the dorsal-ventral polarity of the embryo comes from experiments in which mirr is ectopically expressed in the posterior of the egg chamber. Ectopic expression of DERAF in this region causes the loss of both posterior embryonic segments and the posterior twi expression domain. When mirr is ectopically expressed in the posterior follicle cells, a similar phenotype is observed, with both posterior embryonic denticle belts and the twi expression domain missing. This clearly shows that mirr function is critical for the dorsal-ventral axis of the embryo and that ectopic expression of mirr can lead to an abnormal localization of the ventral signal, which is required to trigger the initiation of the embryonic ventral pattern (Zhao, 2000).

One of the dorsal group of maternal genes, pipe, is negatively regulated by mirr since pipe expression is expanded in mirr minus clones. Since pipe is required for the activation of twi via Dorsal, it is likely that mirr affects twi expression by repressing the expression of pipe in dorsal follicle cells. The downstream targets of mirr, in addition to pipe, need to be further identified to understand how mirr executes these developmental decisions in response to Grk/Egfr signaling. It is possible that mirr is required for the repression of pipe and windbeutel (wind) along with fng in the anterior-dorsal follicle cells. The ventrally-localised activities of these genes then cooperatively generate an extracellular ventral signal. There is evidence that CF2, which is expressed in the ventral and posterior cells in a pattern similar to fringe, regulates pipe and wind, and is responsible for the ventral signal produced in follicle cells that determines the embryonic axis. One might speculate that mirr could repress CF2 in anterior/dorsal follicle cells. However, it is observed that CF2 transcripts are expressed in the mirr expressing cells, and the ventrally-localized protein distribution of CF2 is translationally regulated. Since mirr encodes a transcription factor it cannot be directly responsible for the lack of CF2 protein in anterior/dorsal follicle cells, but could indirectly cause a repression of its translation in the anterior-dorsal cells via another gene. Alternatively CF2 protein could be upstream of mirror expression and repress transcription in ventral cells (Zhao, 2000).

When the mirr mutant phenotype was examined, some unusual results were found, namely that one or two mirr alleles affect only the eggshell, but not the embryonic cuticle, when heterozygous or heteroallelic. All mirr mutant alleles are homozygous lethal, therefore homozygous mutant adults cannot be obtained for analysis. One of the mirr alleles (mirrP1), when heterozygous, balanced with TM3 (mirrP1/TM3), lays either eggs with no dorsal appendages or with a pair of very tiny dorsal appendages, thus showing that some mirr alleles can have either dominant effects on the egg-shell or they interact with other genes in this genetic background. Since this phenotype is not observed in mirror/+, the latter is more likely. Other stocks carrying this TM3 balancer do not show this phenotype under similar conditions, and TM3/TM6 females also lay eggs with normal chorions. Regardless of how this is controlled, the heterozygous mirrP1/TM3 females show reduced expression of mirr and expanded expression of fng in some egg chambers at stage 10 of oogenesis, and lay eggs showing the normal expression pattern of twi. The resulting embryos have normal denticle patterns, and 99% hatch. Thus the mirr P1 allele in a TM3 background affects epithelial morphogenesis, but not embryogenesis. To test further the effects of mirr mutants on eggshell and embryonic development, different genetic combinations of mirr mutants were investigated. Some heteroallelic combinations are lethal, but others generate some adult females. It is suggested that the effect of mirr on the eggshell and embryo can be separated and this either involves mirr functioning separately in two different genetic pathways, and/or different sensitivities to the level of mirr expression for these two functions. The possibility that there are background mutations in the mirr stocks causing the eggshell phenotype cannot be ruled out. However, the possibility that mirr has more than one function is supported by the recent finding that mirr regulates equator formation in eye development by two mechanisms; creating a boundary of fng expression, and reducing the mixing of dorsal and ventral cells at the equator (Zhao, 2000).

In summary, mirr functions as a transcription factor linking the Grk/Egfr signaling in oogenesis to the formation of the dorsal-ventral axis of the egg chamber and eggshell, by modulating the Fringe/Notch/Delta pathway, and by affecting the establishment of the signal from ventral cells to set up the embryonic dorsal-ventral axis. mirr expression in anterior-dorsal follicle cells is activated by Grk/Egfr signaling at stage 9-10 of oogenesis. It then represses the expression in dorsal follicle cells of those genes that are required only in ventral follicle cells. The activity of these genes in the ventral follicle cells are required either for the formation of the eggshell and its dorsal-ventral pattern, or for the initiation of dorsal-ventral axis formation in embryogenesis (Zhao, 2000).


araucan and caupolican are transcribed in the same direction on the chromosome and are separated by 12 kb. mirror is located in chromosomal region 69D, proximal to caupolican, that is, closer to the centromere (Gómez-Skarmeta, 1996).

Genomic size - 23 kb

Transcript size - 3.5 kb

Exons - 6


Amino Acids - 641

Structural Domains

The homeodomains most similar to those of Mirror, Araucan and Caupolican have been found in the human sequence R46202 and in several mouse proteins. The next most similar homeodomains, those of the human Pbx and Drosophila Exd proteins, have only 37% identity. The Mrr homeodomain contains 3 additional amino acids between helix 1 and 2 that are not present in most homeodomains but are present in PBX1 and Exd. Mirror, Ara, and Caup homeodomains are nearly identical. The homeodomain of Drosophila Mirror (McNeill, 1997) has 57 out of 60 amino acids identical to the homeodomains of Caup and Ara. In the amino-terminal domain, these proteins have a highly conserved Notch interaction domain, which has been proposed to be involved in protein-protein interactions. The Notch interaction domain of the homeodomain proteins is homologous to a stretch of amino acids in Xenopus, rat and Drosophila Notch. This putative protein interaction domain is similar to the central part of the epidermal growth factor repeats of the Notch protein. C-terminal to the novel homology region are two potential phosphorylation sites for mitogen-activated kinase (Rolled in Drosophila). Mirror also possesses an acidic activation domain just C-terminal to the homeodomain (Gómez-Skarmeta, 1996 and McNeill, 1997).

mirror: | Transcriptional Regulation | Targets of Activity | Developmental Biology | Effects of Mutation | References

date revised: 26 MAY 97 

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.