The Transcription Factor Optomotor-Blind Antagonizes Drosophila Haltere Growth by Repressing Decapentaplegic and Hedgehog Targets

In Drosophila, decapentaplegic, which codes for a secreted signaling molecule, is activated by the Hedgehog signaling pathway at the anteroposterior compartment border of the two dorsal primordia; the wing and the haltere imaginal discs. In the wing disc, Decapentaplegic and Hedgehog signaling targets are implicated in cell proliferation and cell survival. However, most of their known targets in the wing disc are not expressed in the haltere disc due to their repression by the Hox gene Ultrabithorax. The T-box gene optomotor-blind escapes this repression in the haltere disc, and therefore is expressed in both the haltere and wing discs. Optomotor-blind is a major player during wing development and its function has been intensely investigated in this tissue, however, its role in haltere development has not been reported so far. Here we show that Optomotor-blind function in the haltere disc differs from that in the wing disc. Unlike its role in the wing, Optomotor-blind does not prevent apoptosis in the haltere but rather limits growth by repressing several Decapentaplegic and Hedgehog targets involved both in wing proliferation and in modulating the spread of morphogens similar to Ultrabithorax function but without disturbing Ultrabithorax expression.


Introduction
Drosophila wings and halteres are homologous dorsal structures located in the second and third thoracic segments, respectively. In the larva, the primordia of wing and haltere are sacs of epithelial sheets called imaginal discs. Both structures differ in shape and size; however they share a common developmental program. Indeed, morphogens involved in developmental programs are expressed in a similar fashion in both discs. The acquisition of wing vs. haltere-specific features is controlled by the differential expression of the Hox gene, Ultrabithorax (Ubx), which is expressed in the haltere disc but absent in the wing disc [1,2]. Ubx acts as a haltere identity selector, by impeding the activation of several downstream target genes which are part of signaling pathways common to both wing and haltere discs and by restricting the spread of morphogens within the disc [3][4][5].
optomotor-blind (omb, bifid-FlyBase) is one of the few Decapentaplegic (Dpp; a TGF-β family member) pathway targets [6] that is not repressed by Ubx in the haltere. Omb belongs to the Tbx protein family, which contains a characteristic DNA-binding domain [7], the T-box. Several developmental studies on this protein revealed that Omb regulates wing growth [6,[8][9][10][11] by preserving the wing disc from an excessive apoptosis [8,9]. In the present study we report that, unlike its activities in wing development, Omb antagonizes haltere growth. Our data suggest that Omb controls haltere size by regulating Dpp and Hedgehog (Hh) signal dispersion and by repressing several morphogen targets implicated in wing growth. Importantly, Omb does not seem to regulate Ubx expression in the haltere, which suggests that some of the Dpp and Hh targets are repressed by both Omb and Ubx, either cooperatively or independently.

Transgene overexpression
Ectopic expression of omb was induced in the dorsal compartment of ap-Gal4; UAS-omb/tub-Gal80 ts larvae by maintaining them at 17°C until early third instar (L3) before a shift to 29°C for 48h to inactivate the tubGal80 ts .

Results and Discussion
Sal impedes Dpp-mediated activation of omb in the proximal wing and haltere discs In the wing imaginal disc (Fig. 1A shows wing disc organization), omb is expressed in a large domain around the anteroposterior (A/P) axis in the wing pouch and the presumptive hinge (Fig. 1B, B'). Both Dpp and Wingless (Wg) signaling pathways (see Fig. 1B', B" for the ligands expression) activate omb expression in the wing pouch, whereas only Dpp signaling is required to activate its expression in the prospective hinge [6]. dpp is expressed along the whole A/P boundary in the wing disc; however, omb is not expressed in the proximal part of the disc, the notum (Fig. 1B'). This expression pattern presumes the presence of a repressor, which would prevent Dpp signaling to activate omb in these cells. The gene spalt (sal) is expressed in two broad domains in the wing disc ( Fig. 1D)-one Dpp-dependent in the wing pouch, and another Dpp-independent in the notum [29]. As Sal can act as transcriptional repressor [30], we wondered whether it could be a repressor of omb expression in the notum. Indeed, loss-offunction clones of sal located in the notum relieve the repression (therefore inducing expression) of omb at the A/P compartment border (Fig. 1E), while clones located far from the A/P border do not induce omb expression (Fig. 1E'). Also, double mutant clones for sal and the type I receptor of the Dpp pathway, thickvein (tkv), do not shown an ectopic expression of omb (Fig. 1F), indicating that Dpp signaling is required to activate omb all along the A/P axis but its effect is masked in the proximal wing disc because of the presence of Sal. Note that in Fig. 1E, the derepression of omb within the sal mutant cells does not occurs within the whole clone, indicating that other factor(s) may be involved in omb repression in the notum.
In the haltere disc (see Fig. 1A for haltere disc organization), omb and dpp have a similar expression pattern than in the wing disc (compare Fig. 1B-B" to G-G"). The expression pattern of wg is slightly different between both structures; it is not expressed in the P compartment of the haltere disc (compare Fig. 1B'-B" to G'-G") because of a repression by Ubx [3]. Also, Ubx protein impedes Dpp signaling to activate the distal expression of sal [3], which gets restricted to the proximal part (Fig. 1I). The wild-type expression pattern of Ubx is shown in Fig. 1C and 1H. As in the wing disc, mutant clones for sal in the haltere disc located close to the A/P compartment border derepress omb in the notum (Fig. 1J) while those induced far from the A/P axis do not (Fig. 1J'). Also, double mutant clones for sal and tkv in the proximal part of the haltere abolish the ectopic expression of omb (Fig. 1K). Altogether these results indicate that Sal impedes Dpp signaling to activate omb expression in the proximal part in both dorsal discs. This effect is independent of the Ubx input that each structure received differentially.
Omb prevents haltere growth and does not inhibit JNK-mediated apoptosis in the haltere disc Strong loss-of-function omb alleles are lethal at the pupal stage [6]. Thus, to study adult phenotypes, we removed omb function in mitotic recombinant clones. Wings containing big omb mutant clones lack the central region (compare Fig. 2A and A') [6]. This phenotype is due to Jun N-terminal protein kinase (JNK)-mediated apoptosis [8]. Indeed, puckered (puc)-lacZ expression, a reporter for the JNK pathway, is activated in omb mutant wing discs (compare Fig. 2B and B') [8], and the active form of Caspase-3, a component of the apoptotic pathway, is detected in those discs (compare Fig. 2C and C') [8].
To test whether Omb has a similar role in the haltere, we first analyzed the adult haltere phenotype after eliminating omb function in mutant clones. Because of the relative small size of the haltere, we increased the probability of getting mutant cells in the omb expressing territory by driving the FLP expression in the ptc-Gal4 domain, which is expressed in a stripe of cells at the A/P compartment border. Halteres with such clones do not show lack of tissue. Rather, they present haltere-like tissue overgrowths (compare Fig. 2D and D'; arrowheads) in 17% of the flies observed (n>250). We also induced omb mutant clones by expressing FLP under the control of the Ubx-Gal4 M1 driver. Under this condition, 23% of the flies observed (n>300) have overgrowths in the haltere, which can reach the size of a normal haltere (Fig. 2D"; arrowheads). Ubx-Gal4 M1 flies carry a mutation in Ubx [15] and could also contribute to these overgrowths. To test this hypothesis, we induced omb clones in a background containing a duplication for Ubx. In this case, the overgrowths were smaller but still observed (Fig. 2D"'; arrowhead). This result indicates that omb mutant cells respond differentially depending on Ubx levels.
We then analyzed the effect of Omb on the apoptotic pathway in the haltere disc. Almost no activation of the JNK pathway was detected in omb mutant haltere discs (compare Fig. 2E and E'). Consistently, no significant apoptosis, detected by the activation of Caspase-3, was induced in those omb mutant discs (compare Fig. 2F and F'). Overall, these results reveal a divergent Omb function between the wing and the haltere: contrary to the wing, Omb represses tissue growth in the haltere and does not inhibit JNK-mediated apoptosis in the haltere disc.

In the haltere disc, Omb represses various genes involved in wing growth
We hypothesized that the overgrowths of cells in the haltere triggered by the removal of omb function could be due to alteration of the downstream targets of the signaling pathways normally involved in wing growth, and known to be repressed by Ubx in the haltere [3,31]. Hh signaling is involved in wing growth via the activation, among other targets, of dpp (Fig. 3A), the iroquois (iro) complex genes (Fig. 3B) and the transcription factor collier/knot (col; Fig. 3C) [24,32,33]. In the haltere disc, dpp is expressed in a similar domain than in the wing disc but at lower level (Fig. 3A'), while iro and col expression (Fig. 3B', C') is missing in the haltere pouch disc because of a downregulation or a repression by Ubx respectively [4,34,35]. In omb mutant clones, we observed an upregulation of dpp expression (Fig. 3A"; arrowhead). iro expression was not induced in similar clones (Fig. 3B"). However, a derepression of col expression in omb mutant cells located close to the A/P compartment border was found (Fig. 3C"; arrowhead), coinciding with the homologous region where col expression is induced by Hh in the wing disc.
It has been also reported that Ubx modulates morphogen signaling in the haltere through transcriptional regulation of the two glypicans Division abnormally delayed (Dally) and Dally- like (Dlp) [4,5,31,36]. Dlp and Dally play cooperative and distinct roles in modulating Hh and Dpp gradient and signaling [37]. Both proteins are present in the haltere disc but at a low level in comparison with their expression level in the wing disc (compare Fig. 3D, E to D', E' respectively) due to Ubx in the haltere [4,5,36]. In support of the new role of Omb in the haltere, both glypicans were upregulated in omb mutant clones induced in the haltere disc (Fig. 3D", E"; arrowheads).
In the wing pouch, in addition to omb, Dpp pathway activates vestigial (vg)-QE ( Fig. 4A; visualized with the 5XQE-DsRed reporter), which is known to be involved in wing growth and sufficient to promote wing identity [38], the T-box Dorsocross (Doc)2 (Fig. 4B) [39], which is involved in proximal wing pouch growth [40] and sal (Fig. 4C), which stimulate cell proliferation [41]. None of these Dpp signaling targets is expressed in a wild-type haltere pouch (Fig. 4A', B' and C'). Despite the fact that, vg-QE expression is not induced in omb mutant clones in the haltere pouch ( Fig. 4A"), we were able to detect ectopic expression of Doc2 and sal in similar clones located close to the A/P compartment border in a region homologous to their normal expression domains in the wing (Fig. 4B" and 4C" respectively; arrowheads). The level of derepression of Doc2 protein in those clones appears to reach its endogenous level of expression in the wing disc, suggesting a full derepression. However, the level of expression of sal in omb mutant clones was weaker than its endogenous level in the wing disc. We wondered whether this latter effect could be mediated by the misexpression of brinker (brk), which code for a general repressor of Dpp-dependent genes [42][43][44], expressed normally in the lateral part of both wing and haltere discs (Fig. 4D, D' respectively). However, no appreciable change in brk expression was observed (Fig. 4D"), indicating that another factor would impede a full derepression of sal. Overall, these results suggest that the haltere overgrowths observed upon removal of omb function could be promoted by the derepression of col, Doc2 and sal, genes involved in wing proliferation, and by dally and dlp which are implicated in the modulation of morphogen dispersion and activity. On the contrary, the lack of derepression of vg and iro, both required for wing identity, is consistent with the fact that the adult halteres, where omb mutant clones have been induced, do not present a transformation to wing tissue or a vein pattern.

Omb does not regulate Ubx expression levels
As Ubx represses several Hh and Dpp targets in the haltere [3,31], we wondered whether the derepression of the targets observed in omb mutant clones could be mediated by an alteration of Ubx expression. No change of the level of Ubx protein expression was detected in omb mutant cells ( Fig. 5A; arrowhead). Conversely, the ectopic expression of omb in the dorsal compartment of the haltere disc is not sufficient to affect Ubx expression (Fig. 5B). Interestingly, we note that those discs present a reduction of the dorsal compartment compared to wild-type discs ( Fig. 5B and not shown). In agreement with this observation, most of the adult flies present a reduction of haltere size and some are even devoid of the appendage (compare Fig. 5C to 5C' and not shown); this represents the opposite phenotype of the overgrowths observed in omb mutant clones in adult halteres (Fig. 2D', D").

Conclusions
Our results indicate that Omb restricts haltere growth by repressing the activity of Dpp and Hh signaling which is probably coupled with a control of ligands spreading within the disc. In addition, these effects do not seem to be mediated by changes in Ubx expression, however it is likely that Ubx limits the response of the removal of omb, which could explain why sal is only partially derepressed in those cells. Further experiments will be needed to understand, at a molecular level, how Omb represses Ubx targets in the haltere. It is known that Hox proteins do not act alone but bind cofactors, which confer them specificity in the DNA binding and in gene expression regulation [45]. Therefore, it is possible that Ubx requires Omb as a cofactor to specifically repress a subset of Hh and Dpp targets. Alternatively, Ubx and Omb may act in parallel in this repressive process.