Optomotor-Blind Negatively Regulates Drosophila Eye Development by Blocking Jak/STAT Signaling

Organ formation requires a delicate balance of positive and negative regulators. In Drosophila eye development, wingless (wg) is expressed at the lateral margins of the eye disc and serves to block retinal development. The T-box gene optomotor-blind (omb) is expressed in a similar pattern and is regulated by Wg. Omb mediates part of Wg activity in blocking eye development. Omb exerts its function primarily by blocking cell proliferation. These effects occur predominantly in the ventral margin. Our results suggest that the primary effect of Omb is the blocking of Jak/STAT signaling by repressing transcription of upd which encodes the Jak receptor ligand Unpaired.


Introduction
The Drosophila compound eye originates from the eye-antenna anlage in the embryo. These cells proliferate and form the eye-antennal disc in the larva. In the mid-third instar eye disc, a wave of cell cycle coordination and apical cellular constriction, called the morphogenetic furrow (MF) forms at the posterior margin and progressively moves toward anterior. Posterior to the MF, retinal cell fates are specified by a series of cellular interactions [1,2,3,4]. The early steps of eye development involve at least three aspects: specification of eye fate, control of cell proliferation, and initiation and progression of the MF.
A large number of genes are involved in promoting eye development. Eye fate is specified by the retinal determination gene network which includes the transcription factors encoded by eyeless (ey), twin of eyeless (toy), sine oculis (so), eyes absent (eya), and dachshund (dac) [5,6].
However, developmental processes rarely proceed by agonistic action alone but tend to be held in check by interaction between agonists and antagonists. The necessity to keep retinal development in bounds is obvious in the eye-antennal imaginal disc since this disc, in addition to the retina, gives rise to much of the exterior of the adult head [25,26,27]. Molecules with the ability to block eye development include Patched (Ptc) and other negative regulators of Hh signaling, Wingless (Wg) and the positive components of its signaling pathway, the transcription factors and cofactors encoded by homothorax (hth), teashirt (tsh), hairy (h), extramacrochaetae (emc), pannier (pnr), Chip, arrowhead (awh) and Lim1 [6,13,28,29,30,31,32,33,34].
Of these anti-retinal genes, Wg is the only signaling ligand and appears to be the most important anti-retinal factor. In the third instar eye disc, wg is expressed in the lateral margins and prevents inappropriate marginal morphogenetic furrow initiation [30,35]. Wg exerts its anti-retinal function by several routes. First, Wg blocks MF initiation [30,35]. A primary target is Dpp, which is essential for MF initiation [15,36]. Wg signaling represses dpp transcription and Dpp signaling at a step downstream of receptor activation [37,38]. Second, Wg also blocks MF progression [35] and neuronal differentiation through repression of Daughterless (Da) [38].
Which gene is induced by Wg to mediate its anti-retinal functions? One prime candidate is optomotor-blind (omb, FlyBase bifid, bi) which is expressed in the lateral margins in a pattern similar to the wg expression domain [39]. Ectopic expression of either wg or its downstream effector armadillo (arm) induces the expression of omb near the lateral margins [40]. Omb encodes a T-domain transcription factor and is required for the development of the optic lobes, wing, abdomen, and terminalia [41,42,43,44,45,46,47,48,49,50,51]. The polar eye disc expression and the fact that ectopic omb can completely block eye development [52] led us to investigate the role of omb in this process, and its relationship with Wg.
We show that Omb antagonizes eye development primarily at the level of cell proliferation. We further identified a molecular pathway downregulated by Omb. Our results suggest that the main effects of Omb are a block of Jak/STAT signaling by suppressing transcription of upd encoding the Jak/STAT ligand Unpaired. The block of Jak/STAT signaling accounts for the effect of Omb on cell proliferation. Our results also show that Omb mediates part of the Wg anti-retinal effects.

RNA in situ hybridization
upd RNA in situ hybridization is executed as described [9].

Tissue sections
Semi-thin plastic eye sections were performed according to [69].

Scanning electron microscopy and determination of ommatidial number
Scanning electron micrographs of adult eyes were obtained as described [52]. Due to the curvature of the eye, the ommatidial number N cannot be obtained from a single micrograph. A given eye was photographed from different angles. Dust particles or aberrations in the bristle pattern allowed the alignment of the otherwise repetitive structures such that error-free counting was possible.

omb negatively regulates eye size
In the eye imaginal disc, Omb is expressed in two cell types. Within the main epithelium, Omb is expressed at the dorsal and ventral margins (S1A Fig., arrow). Omb is also expressed in the retinal basal glial cells that lie at the basal level of the eye disc and in the optic stalk [39,70] (S1A Fig.). Only the epithelial expression will be considered here. We found that loss-of-function and gain-of-function omb mutations caused changes in eye size.
In omb hypomorphic allele combinations and omb knock-down, the Omb level was reduced in both margins and the eye disc was enlarged (S1B Fig.). This was observed for the omb hypomorph omb bi in combination with any of three molecularly characterized omb null alleles, l(1) omb D4 , l(1)omb 282 , and l(1)omb 3198 [61,62]. In these adults, the eye was enlarged with an increase in ommatidial number (N) by up to 25%, from 750-800 in wild type to 850-1000 in the mutant (wild type: N = 782, SD = 11.5, n = 6; omb bi /l(1)omb 282 : N = 952, SD = 61, n = 9).
The expansion of the eye occurred primarily ventrally. The dorsal-ventral distinction was based on several criteria. Using the enhancer trap insertion omb P1 [63] as marker for dorsal and ventral ommatidia (Fig. 1A), only an increase in the ventral expression domain could be observed in the adult eye of omb hypomorphs (Fig. 1B). In the larval eye disc, the location of the dorsoventral (DV) midline was defined by the location of the optic stalk (S2A-D, Fig., arrowhead), the inversion of ommatidial chirality based on anti-Bar staining [71] (S2A-B Fig.), the ventral-specific fng-lacZ expression [64] (S2C-D Fig.) and the dorsal-specific mirr-lacZ expression (Fig. 1D, E). There was an obvious enlargement of the ventral eye disc in l(1)omb D4 / omb P7 hypomorphic larvae ( Fig. 1D-F; S2B, D Fig.). We followed the developmental progress of omb P7 eye discs, based on the number of ommatidial rows, and found that the number of ommatidia in the ventral region was consistently higher relative to that in the dorsal region, which was not different from wild type (S2E Fig.). By all these criteria, an overgrowth of the ventral relative to the dorsal part was evident in the omb hypomorphic mutant eye.
These loss-of-function effects were also observed when omb was knocked down by RNAi. Expressing omb-RNAi [48] in its own expression domain using omb P3 -GAL4 caused a strong reduction of Omb level in the margins of eye disc, the retinal basal glia and in the antenna disc This resulted in a strong overgrowth of eye disc (S1B Fig.) and adult eye (Fig. 1B). When omb-RNAi expression was driven by GMR-GAL4, the size of eye disc and adult eye was normal (not shown). This is consistent with omb expression not overlapping with the activity of the GMR-GAL4 driver, which is restricted to cells posterior to the MF [72].
In contrast to the loss-of-function effects, gain-of-function of omb caused reduction or elimination of the eye. In the regulatory dominant gain-of-function allele omb For [51], Omb was overexpressed in the lateral margins and in the retinal basal glia (S1C Fig.). omb For larvae had smaller eye discs (S1C Fig.) and the adults had a reduced number of ommatidia and a posterior indentation in the eye (wild type: N = 782, SD = 11.5, n = 9; omb For : N = 507, SD = 57.6, n = 6). Targeted mis-expression of omb in the lateral and posterior margins by dpp-GAL4 (dpp>omb) causes a strong reduction or total absence of the adult eye [52]. Specific overexpression of omb at the lateral margins using 30A-GAL4 In summary, the loss and gain-of-function phenotypes of omb indicate that omb is a negative regulator of eye development. The effect is stronger on the ventral side of the eye. omb blocks cell proliferation during eye development Deviations from normal eye size can arise by several mechanisms. Changes in proliferation, cell death, morphogenetic furrow progression or retinal differentiation can all affect eye size. We tested the effect of loss and gain of omb on proliferation and cell death.
In omb hypomorphs, cell cycle activity was increased in the ventral eye disc, as monitored by histone H3 phosphorylation (pH3) (omb P7 /Y, Fig. 2B, compare with wild type eye disc in 2A; quantified comparison in 2E) or BrdU incorporation (omb P7 /Y, Fig. 2D, compare with wild type eye disc in 2C) as markers for cell proliferation. There are two mitotic waves in the eye disc. The first mitotic wave occurs anterior to the MF and affects the cell population which will be recruited to form ommatidial clusters. Changing cell proliferation in the first mitotic wave will change the number of ommatidia [9,10]. omb hypomorphic eye discs showed an increase of mitosis in the first mitotic wave (Fig. 2B, D), as expected by their increase in ommatidial numbers. In contrast, the second mitotic wave occurs behind the MF and affects the number of cellular components assembled into the ommatidia [74]. Since omb is expressed in the anterior lateral margins, it is not expected to affect the second mitotic wave. This was confirmed in omb hypomorphic mutants (Fig. 2B, D) and by knocking down omb by GMR>omb-RNAi, which resulted in normally sized adult eyes (not shown). Therefore, omb appears to affect cell divisions in the undifferentiated region anterior to the MF.
We also analyzed the effect of omb loss of function mutant clones on cell proliferation. The l(1)omb D4 null allele was used because it yielded stronger effects than the hypomorphic alleles In wild type eye discs, control clones (marked by lack of GFP) were of similar size relative to their twin spots, irrespective of location (Fig. 2F, G shows two different focal planes to allow clone size visualization and measurement in the infolded margins, in particular the "ventral flap"; data are summarized in Fig. 2J). As expected from the restricted omb expression pattern and from the phenotype of omb loss-of-function mutants, omb null mutant clones had a proliferative advantage relative to their twin spots (omb + /omb + ) only in the ventral margin (Fig. 2I, arrow) but not in the center of the disc (Fig. 2H, arrow) or in the dorsal margin (Fig. 2I). omb clones in the ventral regions were about 3.5 times larger than omb clones in the central region of the disc or than wild type clones (summarized in Fig. 2J).
We next analyzed whether apoptosis plays a role in the omb over-expression phenotype. There is little cell death in wild type larval eye discs [75]. Before onset of retinal differentiation, dpp-GAL4 c 40.6 is expressed in the lateral and posterior margins; later it is restricted to the lateral eye disc margins. The expression in the lateral margins partially overlaps with the omb expression domain in the progenitor region (cf. S6 Fig.). Expression of omb driven by dpp-GAL4 (dpp>omb) caused a strong reduction to total absence of the adult eye [52] and lack of retinal differentiation in the eye disc Enhanced apoptosis could be detected in the posterior margin of the dpp>omb+GFP eye disc (S3A. Fig.). Coexpressing the anti-apoptotic factor p35 (dpp>omb +p35) did not rescue adult eye size (data not shown) nor retinal differentiation in eye disc, although apoptosis was strongly reduced (S3B Fig.). These results suggest that apoptosis is not primarily responsible for eye size reduction at the larval and adult stages and that Omb mainly affects eye size by blocking cell proliferation.

Omb can block retinal differentiation
In addition to the effect on cell proliferation, Omb ectopic expression can block retinal differentiation. Ectopic clonal omb expression at the posterior margin prevented MF initiation (Fig. 3A). Ectopic clonal omb expression in the path of the MF blocked its progression (Fig. 3B) and neural differentiation (Fig. 3C-C-1", arrow). Transient overexpression of omb by heat-induced expression of hs-omb in the entire eye field caused a dorso-ventral scar in the adult eye   (Fig. 3D, arrow) characteristic of furrow-stop mutations [19]. Anterior and posterior to the scar, retinal differentiation proceeded normally, indicating that Omb does not irreversibly arrest MF progression and retinal differentiation. Previously we have shown that sustained omb expression posterior to the MF severely disturbs ommatidial development [52].
Increase in ommatidial number in omb hypomorphs apparently did not occur at the expense of gena tissue (the rim of naked tissue between retina and vibrissae) (Fig. 1B, C). This indicates that Omb in the lateral margins does not act to prevent spreading of eye fate into adjacent tissue domains. Rather, the increase in eye size in omb hypomorphs appears caused by overproliferation of retinal precursors in the ventral eye field (s. above).

Omb inhibits cell proliferation by blocking Jak/STAT signaling through repression of upd transcription
To understand the mechanism by which Omb impedes proliferation in the ventral anterior eye disc, we tested the effect of Omb on the Upd/Jak/STAT signaling pathway, which promotes cell proliferation [7,9,10,18,24], as well as MF initiation [15,17,18,22,24].
The ligand Unpaired (Upd, FlyBase: outstretched, os) of the Jak/STAT pathway is expressed in the ventral eye disc at first instar and in the posterior center of the eye disc at second and early third instar [9,10,57]. Upd, acting through the Jak/STAT signaling pathway, promotes cell proliferation and represses wg transcription to promote MF initiation [7,9,10,18,24]. STAT signaling is induced by Upd and can be detected using STAT reporters containing STAT binding sites [9,10,18,24,76]. Grh-STAT-lacZ and 10X-STAT-GFP reporter expression is high in the posterior region, consistent with STAT activity being induced by the Upd ligand [24]. 10X-STAT-GFP is expressed in the posterior part of the second instar eye disc, before MF initiation. In the third instar eye disc, the 10XSTAT-GFP signal is much reduced and represents perdurance from earlier expression [76]. In the omb hypomorph omb P7 , STAT activity was ectopically activated in the ventral margin ( Fig. 4B and S4 Fig.), as monitored by expression of 10XSTAT-GFP-nls (constructed in this study) which is normally expressed only in the posterior region of the eye disc (Fig. 4A). omb P3 >omb-RNAi yielded similar results (not shown). A STAT-lacZ enhancer trap reporter, although not fully recapitulating the STAT92E mRNA pattern, is known to be negatively regulated by Jak/STAT activity [43,57] and, therefore, can be used as a reporter for Jak/STAT activity. In the wild type L3 eye disc, its expression was higher in the lateral poles and lower around the DV midline (Fig. 4C, see also [57]). In the l(1)omb 15 eye disc, STAT-lacZ expression was lost in the ventral region (Fig. 4D), suggesting an elevated STAT activity in this region. To determine whether omb regulates Jak/STAT activity cell-autonomously, we generated l(1)omb D4 mutant clones. We found that 10XSTAT-GFP-nls was nonautonomously induced in ventral clones (Fig. 4E-E"'). These results suggest that Omb normally acts to repress Jak/STAT activity in the ventral region of the eye disc.
In order to determine, at which level of the Jak/STAT signaling cascade Omb inhibits this pathway, we coexpressed the Janus kinase gene hopscotch (hop) with omb (dpp>omb+hop) (Fig. 4G). Hop largely rescued retinal development indicating that Omb acts upstream of hop. dpp>hop (not shown) and dpp>hop+GFP (Fig. 4F) caused an enlargement of the eye disc, consistent with the role of Jak/STAT signaling in promoting cell proliferation. These results suggest that repression of Jak/STAT activity is a major mechanism by which ectopic Omb blocks eye development. We then asked whether increased Jak/STAT signaling can account for the omb loss-of-function phenotype. We reduced the STAT92E dose in the omb P7 mutant background, (omb P7 /Y; STAT 06346 /+ and omb P7 ; STAT92E 397 /+), which caused attenuation of the ventral outgrowth of eye disc (Fig. 4H, H', two focal planes; quantified comparison in 4I). These results suggest that the overgrowth phenotype elicited by reduced omb expression requires Jak/STAT activity. Therefore, in its normal function Omb appears to suppress inappropriate Jak/STAT signaling at the ventral margin.
Upd is the ligand of Jak/STAT pathway and expressed in the center of the posterior margin of the eye disc in L2 and L3 larvae (S5A-D, F Fig.; [9,57]). We examined whether omb could suppress Jak/STAT activity by downregulating upd expression. Expression of omb along the posterior margin (in dpp>omb+GFP) suppressed upd-lacZ expression (S5E, E' Fig.). Clonal expression of omb repressed upd-lacZ cell autonomously (Fig. 5A-A"'). We further tested whether upd expression is affected in omb mutant eye discs. We performed RNA in situ hybridization on wild type and omb P7 eye discs and found that upd was derepressed in the ventral margin in omb P7 (Fig. 5C) compared wild type (Fig. 5B). Thus, ectopically expressed Omb can suppress upd transcription, and Omb in its normal expression domain restricts upd transcription in the ventral eye margin. Moreover, coexpression of omb with upd (in dpp>omb +upd) largely rescued retinal development in the eye disc (Fig. 5D) and the adult eye (Fig. 5E). These results suggest that Omb, in its endogenous expression domain, acts by repressing upd, thus limiting cell proliferation in the ventral eye. Our results further suggest that repression of Jak/STAT signaling occurs at the transcriptional level of upd and is the major mechanism by which Omb blocks cell proliferation in eye development.

Omb is a mediator of Wg anti-retinal function
Since Wg and Omb are expressed in a similar pattern and both block retinal development, we asked whether the anti-retinal activity of the Wg signal is mediated by Omb. The Wg effector Arm was expressed to mimic Wg signaling. In dpp>arm+GFP, eye disc size was reduced and no neuronal differentiation was detected (Fig. 6A), consistent with a block of eye development by Wg signaling. dpp expression along the posterior and lateral margins has little overlap with endogenous wg expression (S6A'-E', A"-E" Fig.). Therefore, this experiment tests the effect of ectopic Wg signaling. When omb was knocked down in the background of dpp>arm (in dpp>arm+omb-RNAi), disc size and neuronal differentiation were partly recovered (Fig. 6B). Adult eye size also was largely restored (Fig. 6D). Knock down of Omb in the dpp expression domain (dpp>omb-RNAi+GFP) caused no significant effect on retinal development (Fig. 6C, E). Misexpression of arm by ey-GAL4 (ey>arm) caused a reduction of adult eye size and retinal differentiation in eye disc (Fig. 6F, J). When omb dosage was reduced (l(1)omb D4 /+), the ey>arm phenotype was partially rescued with full penetrance (Fig. 6G-I, K-L). These results suggest that Omb is one of the mediators of Wg activity in blocking eye development.

omb represses retinal development
In this study, we demonstrate that omb is a negative regulator of retinal development. Omb can block eye development at several levels: cell proliferation, MF initiation, and progression. In omb loss-of-function mutant or RNAi-knockdown animals, the most prominent phenotype was an enlargement of the ventral eye (Fig. 1) due to extra cell proliferation (Fig. 2). omb mutant clones in its expression domain at the ventral margin were 3.5 times larger than control clones (Fig. 2). These loss-of-function results were supported by the opposite effect in gainof-function experiments. In omb gain-of-function animals, the size of eye disc (S1C Fig.) and adult eye was reduced (data not shown). In addition, omb-expressing clones blocked MF initiation (Fig. 3A) and progression (Fig. 3B). Expression of omb along the margins (dpp>omb +GFP) could completely block retinal development (S3B-C Fig.).
Act5C>omb clones were not detected in late third instar eye discs. Rare clones were observed only when larvae were raised at 17°C and examined at early to mid-third instar (Fig. 4).
These clones were often round and sorted out from the neuroepithelial layer. This behavior has previously been observed for omb gain-of-function clones in the wing imaginal disc [49].
We identified the Jak/STAT signaling pathway asdownregulated by Omb. In omb lossof-function animals, the 10XSTAT-GFP-nls and upd-lacZ expression were elevated in the ventral region (Figs. 4B, 5C). Conversely, when omb was ectopically expressed in the posterior and lateral margins, upd-lacZ expression at the margins was reduced (S5E, E' Fig.). These results show that upd transcription is repressed by Omb in the lateral margins, especially the ventral margin.
The repression of the Upd signaling cascades by Omb is of developmental relevance. Ectopic expression of omb in the margins (dpp>omb+GFP) blocked retinal development (S3C,5E Figs.). Coexpression with of hop (dpp>omb+hop) could nearly fully restore retinal development in the eye disc (Fig. 4G). Reducing the dosage of STAT (in omb P7 /Y; STAT92E 397 /+) partially suppressed the omb P7 enlarged eye phenotype (Fig. 4H), suggesting that STAT signaling is downstream of omb and involved in causing its mutant phenotype. These results indicate that the repression of the Upd/Jak/STAT pathways is responsible for the block of retinal development by Omb.
upd is transcriptionally repressed by Omb. Is it a direct transcriptional target of Omb? Omb is a transcription factor of the T-box family all members of which bind to a common consensus  upstream region of upd (G. O. Pflugfelder, unpublished data). This finding and the cell-autonomous repression of upd by Omb (Fig. 5A) as well as the derepression of upd in the ventral omb domain in omb mutant eye discs (Fig. 5C) suggest that repression of upd by Omb may be direct. However, Omb cannot be the only factor that prevents upd expression from the lateral margins, because dorsally upd is not derepressed in omb mutant eye discs.
Omb partially mediates the anti-retinal function of Wg signaling Since Wg and Omb are expressed in similar patterns and since both can block retinal development at multiple steps, the question arises whether Omb mediates Wg signaling. We found that the anti-retinal effects of ectopic Wg signaling (ey>arm and dpp>arm) were attenuated when the omb dosage was reduced. The partial rescue suggests that Omb mediates part of the Wg effect and that additional factor(s) are likely to be involved. Our analyses show that the effect of Omb is partly similar to that of Wg and partly different (Fig. 6).
A clear difference was the opposite effect on cell proliferation. Omb repressed cell proliferation, through a block of Upd/Jak/STAT signaling. In contrast, enhanced Wg signaling (in axin mutant) causes overgrowth [78], whereas loss of Wg results in reduction of eye disc size [30,35]. Therefore, Wg can promote cell proliferation.
Loss of omb affected primarily development of the ventral eye margin. The ventral bias included effects on STAT activity. Therefore, the effect of wg in the dorsal side may be mediated by another factor, either independent of Omb or functionally redundant with Omb. Dorsal eye fate is governed by the expression of members of the Iroquois gene complex (Iro-C) [79]. Its activity modulates the function of genes that are symmetrically expressed at the poles of the eye disc, like teashirt, for instance [79,80]. The specification of dorsal rim ommatidia late in eye development provides an example of how Omb function is modulated at the lateral eye margins.
During pupal eye development, a dorsal Wg gradient specifies three cell fates at the dorsal eye boundary: pigment rim, polarization-sensitive dorsal rim (DR) ommatidia, and bald ommatidia (lacking the mechanosensory hairs). Omb which is induced by Wg, is sufficient to induce the DR fate in the dorsal eye. Ectopic omb induces the dorsally restricted expression of homothorax (hth) which together with ubiquitous extradenticle (exd) allows the formation of Hth/Exd complexes which specify DR development [81]. Dorsalisation of the ventral eye by ubiquitous expression of Iro-C genes causes DR fate also along the ventral margin. The monopolar Iro-C expression thus determines the different functional outcome of symmetrical omb expression at the two margins. Like in early eye development, loss of omb has little consequence for dorsal eye development during specification of DR ommatidia. The discrepancy between strong effect in omb gain-of-function and little effect in omb loss-of-function dorsal eye phenotypes has been attributed to a redundant function exerted by the related T-box genes Dorsocross (Doc) [82].
Intriguingly, the closely omb-related vertebrate T-box genes Tbx2/3/5 are all expressed in a polar pattern, first in the dorsal eye cup and later in the dorsal retina [83,84,85,86]. A role of human TBX5 in eye development is apparent from the frequent ophtalmological symptoms of patients suffering from Holt Oram Syndrome (TBX5 haploinsufficiency) [87,88]. The related expression patterns of Tbx2/3/5 and omb in eye development suggest conservation of ancient functions which were already present in an evolutionary precursor before the split into the protostome and deuterostome lineages [89], in spite of the widely differing mechanisms of eye ontogenesis in metazoans [90,91]. The D/V eye field was symmetrical in the third instar eye disc of wild type, but the ventral field was expanded in omb P7 /Y. (E) The number of rows of ommatidia in each eye disc, and the numbers of ommatidia in the dorsal and ventral eye fields were counted at different stages of eye disc development. In wild type eye disc, the dorsal and ventral eye fields were always of equal size. In omb P7 /Y, the ventral eye field was consistently larger than the dorsal field. In early eye disc, dpp-GAL4 expression is similar to that of dpp-lacZ (S6 Fig.) in the posterior and lateral margins. Unlike dpp-lacZ, dpp-GAL4 is not expressed in the progressing MF in mid to late third instar eye disc. (B) dpp>omb+GFP, as dpp>omb, completely blocked eye development in adult (B) and in late third instar eye disc (C). The eye disc has no neuronal differentiation (Elav, blue) but has elevated activated caspase 3 (red). (D) Blocking apoptosis by coexpression of p35 (dpp>omb +p35) significantly reduced the caspase 3 signal but did not rescue eye size or retinal differentiation. Scale bar: 50um. The expression patterns of omb (visualized by omb P3 >GFP, green), (A'-E') Wg (anti-Wg, red), and (A"-E") dpp (represented by dpp-lacZ, blue) were followed during eye-antennal disc development from early second instar to late third instar. (A"'-E"') shows the merge images of omb P3 >GFP, dpp-lacZ and Wg immunostaining. (TIF)