Specific expression and function of the Six3 optix in serially homologous organs

Organ size and pattern results from the integration of two positional information systems. One global, encoded by the Hox genes, links organ type with position along the main body axis. Within specific organs, local information is conveyed by signaling molecules that regulate organ growth and pattern. The mesothoracic (T2) wing and the metathoracic (T3) haltere of Drosophila represent a paradigmatic example of this coordination. The Hox gene Ultrabithorax (Ubx), expressed in the developing T3, selects haltere identity by, among other processes, modulating the production and signaling efficiency of Dpp, a BMP2-like molecule that acts as a major regulator of size and pattern. Still, the mechanisms of the Hox-signal integration even in this well-studied system, are incomplete. Here we have investigated this issue by studying the expression and function of the Six3 transcription factor optix during the development of the Drosophila wing and haltere development. We find that in both organs Dpp defines the expression domain of optix through repression, and that the specific position of this domain in wing and haltere seems to reflect the differential signaling profile among these organs. We show that optix expression in wing and haltere primordia is conserved beyond Drosophila in other higher diptera. Despite the similar expression pattern, optix plays different roles in wing and haltere. In the wing, optix is required for the growth of the most anterior/proximal region (the “marginal cell”) and for the correct formation of sensory structures along the proximal anterior wing margin. In contrast, in the haltere optix is necessary for the suppression of sensory bristles without any noticeable effect on organ growth. Therefore, optix shows an organ-specific function. Beyond dipterans, optix expression in the anterior wing has been shown also in butterflies. We propose that the ancestral role of optix might have been structural in the anterior wing. Once expressed in the wing, optix expression had been re-deployed for wing spot formation in other parts of the wing of Heliconius butterflies.


INTRODUCTION
During development, several positioning systems inform cells of their location. First, the Hox code defines position along the anterior-posterior axis. In insects, this system generates segmental diversity along this body axis. Next, in each segment, cells within an organ primordium obtain positional information from local signaling centers (Mann and Carroll, 2002). The Drosophila wing and haltere primordia constitute a paradigm where the integration of these two positional systems has been intensively investigated. In the Drosophila thorax (T), the second and third segments develop two serially homologous organs: the wing (in T2) and the haltere (in T3). The specific expression of the Hox gene Ultrabithorax (Ubx) in T3 is responsible for the specific morphology of the haltere, a small, modified wing that acts as balancing organ during Drosophila flight. Mutants that cause the loss of Ubx during haltere development cause its transformation into wing (Lewis, 1978), while ectopic expression of Ubx in the developing wing results in wings transformed into haltere-like appendages (Gonzalez-Gaitan et al., 1990).
One of the major organ-positioning systems in the wing and haltere primordia -or "imaginal discs"-is a stripe of cells that bisects the disc and produces a BMP2 ligand, encoded by decapentaplegic (dpp). From this stripe, Dpp generates a signaling gradient that patterns the anterior-posterior axis of the appendage (Restrepo et al., 2014). In the wing, which is the ground state of the dorsal appendage, the Dpp gradient activates the nested expression of a number of target genes at different concentrations, such as spalt (sal, referring collectively to two highly related paralogous genes, sal-m and sal-r) and optomotor blind (omb) (Nellen et al., 1996). sal and omb activation is indirect, though, by Dpp signaling repressing brinker (brk), itself a repressor of the Dpp pathway (Campbell and Tomlinson, 1999;Winter and Campbell, 2004). In this way, Dpp controls the positioning of central pattern elements, such as the wing veins (de Celis et al., 1996;Sturtevant et al., 1997). In addition to patterning, Dpp signaling controls organ growth, so that mutants that lack Dpp signaling result in very reduced winglets (Posakony et al., 1990).
In the haltere, Ubx modifies the wing developmental program in two ways. First, as a transcription factor, Ubx regulates the expression of some targets. For example, Ubx represses sal expression (Weatherbee et al., 1998). Second, Ubx modifies the shape of the Dpp-generated signaling gradient indirectly, by controlling the expression of proteoglycans required for Dpp dispersion (Crickmore and Mann, 2006;de Navas et al., 2006b). Globally, these modifications of Dpp signaling and target gene activation by Ubx have been related to the size and patterning differences between halteres and wings.
Although Dpp signaling generates a signaling gradient that spans the whole wing pouch and its activity is required throughout the wing, so far only centrally expressed target transcription factors have been described. Here we report the functional involvement of the transcription factor Optix/Six3 in patterning of the most anterior region of the wing and the haltere. In both discs, optix expression is anteriorly restricted by Dpp signaling, although in the wing the precise expression boundary may be set with the collaboration of wing specific Dpp targets, such as sal. We show that optix shows organ-specific functions: in the wing, it is necessary for the growth of the anterior/proximal wing ("marginal cell") and the development of wing margin sensory bristles, while in the haltere optix is required for sensory bristle formation.
Overexpression of optix in the entire wing pouch affects only anterior wing development, suggesting that other parts of the wing cannot integrate ectopic Optix input. This observation may provide a mechanistic explanation for a widespread redeployment of optix expression in wing spot formation in various butterfly species.

Fly strains and genetic manipulations
Two optix:GFP lines were examined (Sarov et al., 2016) (318456/88 and 318371/10042). Expression of both lines was qualitatively similar, but the signal of optix:GFP 318371/10042 was strongest and was used for all further studies (and referred to as "optix:GFP").
The UAS/GAL4 system (Brand and Perrimon, 1993) was used for most gain-and lossof-function assays. Given that we focused our investigation in the pouch region, we used as wing specific driver nubbin-GAL4 (nub-GAL4)  To generate tkv loss-of-function clones through mitotic recombination (Xu and Rubin, 1993) we crossed 1096-GAL4,UASflp;FRT40arm-lacZ/CyO females to tkv a12 FRT40A/CyO males. tkv a1 is a tkv-null allele. In this experiment, the bxMS1096-GAL4 line ("1096-GAL4" (Milan et al., 1998) drives UAS-flipase throughout the wing disc to induce mitotic recombination clones in this organ specifically. optix:GFP was introduced in these genotypes by standard genetic techniques. All crosses were raised at 25°C, except in the case of UAS-RNAi experiments, which were transferred to 29°C 48 hours post-fertilization (hpf) to maximize the penetrance of the knockdowns.
The Mediterranean fly strain Ceratitis capitata Egypt II was obtained from the FAO/IAEA Agriculture and Biotechnology Laboratory (Seibersdorf, Vienna, Austria) and reared at 28°C and 55 ± 5% RH. The house fly strain Musca domestica ITA1 was collected in Italy, Altavilla Silentia in 2013 (Y. Wu and L. Beukeboom, GELIFES, The Netherlands) and kept at room temperature (22±2°C) on wheat bran-based food.

in situ hybridization
Images of optix mRNA expression in wing and haltere discs are unpublished data kindly shared by P. Tomancak (MPI-CBG, Dresden) and C. Dahmann (Technische Universität, Dresden), obtained using probes and methods described in (Tomancak et al., 2007 3.5µl BCIP® (5-Bromo-4-chloro-3-indolyl phosphate disodium salt, SIGMA-ALDRICH® Chemie GmbH, Munich, Germany, 50 mg/ml in 100% DMF)). Samples were incubated in the dark at room temperature. The staining reaction was stopped by washing samples three times with PBT for 10 min each.

Immunofluorescence and confocal imaging
Immunofluorescence in wing and haltere imaginal discs was carried out according to standard protocols. Primary antibodies used were: mouse anti-GFP (Molecular Probes), rabbit anti-GFP (Molecular Probes), rabbit anti-pSmad3 (AbCam), anti-Sal (gift of C. Sánchez Higueras and J. Hombría, CABD, Seville) and anti-Ubx (#FP3.38, Iowa University Hybridoma Bank). lacZ reporters were detected using a rabbit anti-βgalactosidase antibody (#55976, Cappel). Appropriate Alexa Fluor-conjugated secondary antibodies were used. For experiments that were used for fluorescence intensity quantification, confocal settings were kept constant, so that fluorescence intensity could be compared across discs. After immunostaining, samples were imaged using a Leica SPE confocal microscope. Images were processed with ImageJ (NIH).

Quantification of gene expression profiles
Expression profiles for optix:GFP, PMad (monitored using a cross-reacting antibody against P-Smad3), sal and omb (monitored in omb>RFP larvae) were obtained using the "Plot profile" function of ImageJ (NIH). Within each experiment, confocal settings were maintained constant so intensity profiles would be comparable. Intensity profiles were expressed in arbitrary units.

RESULTS
The Six3 gene optix is differentially expressed in the wing and haltere discs.
Optix transcription, detected using RNA in situ hybridization, is found in both the wing and haltere imaginal discs of late third instar (L3) larvae (Organista et al., 2015;Seimiya and Gehring, 2000) ( Figure 1A) in the pouch regions of both discs. These pouch regions give rise to the wing proper and the distal haltere's article (capitellum), respectively ( Figure 1C; (Cohen, 1993). To examine the expression of optix in detail, we used an Optix:GFP line (Sarov et al., 2016) that recapitulates optix expression   domestica. In both species, the expression pattern in wings and haltere discs is very similar to the one seen in D. melanogaster.

optix plays different roles in wing and haltere development
In order to determine the role played by optix, we drove an optix-RNAi to the distal wing and haltere discs, using the nubbin-GAL4 (nub>) driver ( Figure 3A). The adult wings of nub-GAL4; UAS-optixRNAi ("nub>optixRNAi") flies were bent and smaller than those of their nub>+ siblings ( Figure 3B). This phenotype seemed to be mostly due to a much shorter longitudinal vein 2 (L2) and a reduction of the wing blade area anterior to this vein (the so-called "marginal cell", in between veins 1 and 2; Figure   In the wing and haltere primordia AP information is conveyed by Dpp (Decapentaplegic), a BMP2-like molecule. Dpp is produced at a stripe along the AP axis from where it diffuses, creating a signaling gradient (Restrepo et al., 2014). Cells regulate gene expression according to the signaling levels they perceive -i.e.
according to their position. The read-out of this Dpp signaling gradient is the phosphorylated form of the transcription factor Mad (pMad) (Sekelsky et al., 1995).
However, the shape of the Dpp signaling gradient (i.e the pMad profile) differs between wing and haltere discs. This difference has been shown to be the result of Ubx regulating the production and spread of Dpp in the haltere (Crickmore and Mann, 2006;de Navas et al., 2006a). In addition, Ubx regulates directly the output of the Dpp pathway, for example, by repressing Dpp's target spalt (sal) in the haltere (Barrio et al., 1999;Galant and Carroll, 2002;Weatherbee et al., 1998). Since we have found that optix was expressed more laterally in the wing than in the haltere pouch, we asked whether the position of optix relative to the Dpp signaling gradient was also different.
To do this, we stained optix:GFP discs for pMad, using a crossreacting antibody against the mammalian pSMAD3. We confirmed previous observations indicating that that the maximum intensity of the pMad gradient was lower in the haltere, indicating a weaker Dpp signaling in the haltere compared to the wing (Figure 4; (Crickmore and Mann, 2006). Within this gradient, optix was excluded from regions of high and intermediate signal (100-600 arbitrary units) in the wing ( Figure 4B) as well as in the haltere, where optix was displaced relatively to the pMad gradient, so that its expression was excluded from the peak of pMad signal (100-150 arbitrary units) ( Figure 4A-D). This meant that the expression domain of optix relative to the Dpp signaling intensity was similar in both discs despite their optix domains being located far from (in the wing) or adjacent to (haltere) the AP border ( Figure 1). The complementarity of expression suggested that the positioning of optix expression was set by Dpp repressing optix.

Dpp signaling represses optix and sets the limits of its expression domain.
Dpp signaling could be repressing optix directly or indirectly, through some of its targets. To test the hypothesis of a repressor role for the Dpp pathway, we first attenuated the expression of the Dpp type II receptor punt (Letsou et al., 1995;Ruberte et al., 1995) using a punt-specific RNAi ( Figure 5). In nub>puntRNAi wing and haltere discs the optix domain extended towards the disc center. In these discs, the pMad signal from the pouch is absent, confirming the blockade of the signaling. This was especially noticeable in the wing disc ( Figure 5A,B). Although this result suggested a repressive action of the Dpp pathway on optix, it could not rule out that low levels of Dpp signaling could be activating optix as, in our experiment, punt levels had been attenuated using an RNAi. To unambiguously assess the role of Dpp in the regulation of optix, we induced loss-of-function clones of tkv, the Dpp type I receptor (Brummel et al., 1994;Nellen et al., 1994). In these clones, that grow poorly and tend to extrude from the epithelium , we detected depression of optix:GFP in clones all along the AP axis of the wing ( Figure 5C). As these clones cannot transduce the Dpp signal, we conclude that Dpp signaling is a repressor of optix. Figure 4. optix expression relative to the pMad signaling profile. Optix:GFP wing (A) and haltere (C) discs stained for pMad (red), and counterstained with the nuclear marker DAPI (blue). GFP and pMad profiles for wing (B) and haltere (D) discs, across the white lines, are shown. Fluorescence intensity in arbitrary units. "A" and "P": "anterior" and "posterior" compartments, respectively. The AP border is marked by the white dashed line.

Figure 5. Dpp signaling represses optix:GFP. nub>optixRNAi; optix:GFP wing (A)
and haltere (B) discs. optix:GFP signal expands. In these discs, the pMad signal characteristic of the pouch is lost. Discs are counterstained with the nuclear marker DAPI. (C) Wing disc containing tkv-mutant clones. The clones are marked by the absence of armZ (C' shows the armZ signal alone). As these clones tend to sort out from the epithelium, they are detected as small "holes" in an apical view. To the right, a detail of the same wing pouch, showing an apical view ("x,y") and a confocal z-optical section ("x,z"). Arrows mark clones in the (x,y) and (x,z) planes. All these clones, which are small and sort out basally, express optix:GFP.
To examine the possibility that the repressive action was exerted through some of its targets, we analyzed the expression of optix in wing and haltere discs relative to two known Dpp target transcription factors, spalt (sal) and optomotorblind (omb) (de Celis et al., 1996;Grimm and Pflugfelder, 1996;Kim et al., 1996;Nellen et al., 1996;Sturtevant et al., 1997). In the wing, the sal and optix domains are separated by an intermediate zone and do not overlap ( Figure 6A,A'). In the haltere pouch, though, sal is not expressed ( Figure 6B,B') and yet, as we showed above, optix expression is excluded from the regions of intermediate/high Dpp signal. These results do not rule out sal as an optix repressor in the wing, but suggest that it cannot be the sole repressor, as it is absent from the haltere. Indeed, Organista et al. (2015) have shown that in sal-mutant wing discs, optix expression extends towards the disc's center, but does not reach the AP boundary, indicating that additional Dpp-dependent mechanisms for optix repression must exist. Next, we analyzed the expression of optix relative to omb in omb-GAL4; UAS-cherry-RFP. While in the wing optix and omb were complementary to one another ( Figure 6C,C'), in the haltere we detected significant overlap between both genes ( Figure 6D,D'). Therefore, omb does not seem to fulfill the repressor role either, because of its coexpression with optix in the haltere. We find that there is a two-fold relationship between optix and Ubx. On the one hand, Ubx is responsible for the modification of Dpp positional system in the haltere, which then sets optix domains along the AP position. On the other, optix expression is required in the haltere to suppress the formation of sensory bristles in this organ -a function known to be exerted by Ubx (Garcia-Bellido and Lewis, 1976;Weatherbee et al., 1998). In principle, this latter phenotype could be produced if optix were required for either Ubx expression or function. Alternatively, optix could be required for one of Ubx's activities: the suppression of bristle development. We tested the first possibility by examining Ubx expression in nub>optixRNAi haltere discs, stained with an anti-Ubx antibody. In these discs we did not observe any change in Ubx protein levels relative to controls ( Figure S2A,B). This result was not unexpected, as a reduction in Ubx levels would have led, in addition to extra bristles, to an increase in haltere size, something we do not observe in nub>optix RNAi individuals. Therefore, we favor the second alternative: that optix is necessary for bristle suppression by Ubx in the haltere.
Forced expression of optix throughout the wing disc results in extravenation in the anterior wing, but does not affect the rest of the organ.
The fact that optix expression was restricted to the anterior-most region of the wing disc made us ask whether optix might affect wing development if ectopically expressed throughout the developing wing. We tested this by driving a UAS-optix transgene using nub-GAL4. Wings of nub>optix adults showed extravenation in the margin cellprecisely the region where optix is normally expressed and required (Figure 7).
However, the rest of the wing remained unaltered. This lack of effect was particularly unexpected: Optix is a transcription factor and we would have predicted that, similarly to what happens in the anterior-most wing, cells elsewhere in the pouch would have responded to its ectopic expression. This to us suggests that most parts of the wing pouch are "protected" from the action of optix, either by lack of available DNA target sequences (an epigenetic effect) or the absence of a positive co-factor (or the presence of a repressor).

DISCUSSION
The Dpp signaling gradient is required for the patterning of the whole wing, from the center to its margin (Lecuit et al., 1996;Nellen et al., 1996;Posakony et al., 1990). This gradient is translated into a series of contiguous domains expressing distinct transcription factors, each required for the specification of specific features in the adult organ (Lawrence and Struhl, 1996;Restrepo et al., 2014). However, while the transcription factors acting in the central wing were known, the most anterior region of the wing -the region comprised between the longitudinal vein 2 (L2) and the anterior margin (L1)-lacked a specific transcription factor. In this paper, we show that this transcription factor, or at least one of them, is optix.
Our results indicate that optix is expressed in, and required for the growth of this most anterior sector of the wing, the so-called margin cell. This role is in agreement with previous results showing that Six3 regulates cell proliferation in vertebrate systems (Del Bene et al., 2004;Gestri et al., 2005). We further show that Dpp signaling plays a major role in setting the optix expression domain. Although it has been reported before that sal genes are required to set the central limit of this domain, in discs lacking sal function optix does not extend all the way to the AP border (Organista et al., 2015), optix in the haltere also seems to exclude omb as a repressor. Therefore, either another unknown repressor exists, or Dpp signaling acts as a direct optix repressor.
While in the haltere, the domain of optix would be set directly by Dpp, in the wing sal would be an additional repressor. By intercalating sal, the Dpp positioning system may be able to push the limit of optix expression farther away from the AP border of the wing. The Sal proteins have been previously shown to act as transcriptional repressors of knirps (kni) to position vein L2 (de Celis and Barrio, 2000). Thus, adding sal repression may help to align the optix domain with L2. This additional repression would not be operating in the haltere, which lacks venation.
Interestingly, the logic of optix regulation by Dpp is different from that of other Dpp targets. The activation of the sal paralogs (sal-m and sal-r) and aristaless (al), another target required for vein L2 formation (Campbell and Tomlinson, 1998), proceeds through a double repression mechanism: In the absence of signal, the Brinker repressor keeps sal and al off. Activation of the pathway leads to the phosphorylation of the nuclear transducer Mad (pMad) which, in turn, represses brk, thus relieving the repression on sal and al (Campbell and Tomlinson, 1999;Moser and Campbell, 2005).
Therefore, optix regulation by Dpp signaling could be more direct similarly to that of brk.
One interesting aspect of optix function is that it plays qualitatively different roles in the wing and the haltere. While in the wing optix is required for the development of the anterior-most portion of the wing (including the margin bristles), in the haltere optix serves to suppress the development of sensory bristles, a task known to be carried out by the Hox gene Ubx. We have ruled out a role for optix in regulating Ubx expression, at least when judged from Ubx protein levels ( Figure S2). Therefore, optix is required for a subset of Ubx's normal functions. Since optix encodes a Six3-type transcription factor, this interaction could be happening at the level of target enhancers, where the combination of Ubx and Optix would allow the activation or repression of specific sets of genes.
Finally, we have observed that the expression of optix in wing and haltere primordia is conserved across higher Diptera ( Figure 2). Interestingly, optix is expressed in the developing wings of passion vine butterflies (genus Heliconius). In Heliconius species, optix has been co-opted for red color patterning in wings (Reed et al., 2011). However, the ancestral pattern found in basal Heliconiini is in the proximal complex, a region that runs along the base of the forewing costa, the most anterior region of the forewing (Martin et al., 2014). This similarity between optix expression patterns in forewings of Diptera and Lepidoptera make us hypothesize that an optix ancestral role might have been "structural", being required for the development of the anterior wing. Once expressed in the wings, recruitment of red pigmentation genes allowed optix co-option for color pattern diversification through regulatory evolution (Martin et al., 2014). We note that a pre-requisite for this co-option in wing pigmentation patterning must have been that optix would not interfere with the developmental pathway leading to the formation of a normal wing in the first place. The fact that the effects of optix overexpression throughout the wing primordium in Drosophila are restricted to the anterior/proximal wing -its normal expression domain-indicates that optix cannot engage in promiscuous gene regulation, and that its function depends on other competence factors, which would limit its gene expression regulatory potential. Figure S1. Attenuation of optix causes the loss of triple row sensory bristles. Control (A: nub>) and optix knock-down (B: nub>optixRNAi). Arrows in (A) and (B) mark equivalent positions. Note absence of bristles along the anterior-proximal margin that spans the reduced marginal cell.