Sensory perception of light is mediated by specialized photoreceptor neurons of the eye. Each photoreceptor expresses unique photopigments called opsins and they are sensitive to particular wavelengths of light. In insects, ocelli and compound eyes are the main photosensory organs and they express different opsins. It is believed that opsins were duplicated during evolution to provide specificity to ocelli and the compound eye and this is corelated with their distinct functions. We show that Homothorax acts to control a binary Rhodopsin switch in the fruit fly Drosophila melanogaster to promote Rhodopsin 2 expression and represses Rhodopsin 1 expression in the ocelli. Genetic and molecular analysis showed that Homothorax acts through the promoters of rhosopsin 1 and rhosopsin 2 and controls their expression in the ocelli. We also show that Hth binding sites in the promoter region of rhodopsin 1 and rhodopsin 2 are conserved between different Drosophila species. We therefore proposed that Hth may have acted as a critical determinant during evolution which was required to provide specificity to the ocelli and compound eye by regulating a binary Rhodopsin switch in the ocelli.
The ability to perceive and discriminate a broad range of environmental stimuli in nature is essential for many aspects of life. Animals rely heavily on visual cues to perform complex tasks such as navigation to find food, mates and shelter as well as social interactions. Visual cues are perceived by visual organs that contain photoreceptors (PRs) as light-sensing structures. PRs are specialized cells that gather information from the surrounding world which is subsequently processed by the brain. Each PR expresses a unique photosensitive opsin/rhodopsin that defines the wavelength of light by which a PR will be activated.
It is believed that eyes have evolved separately due to fundamental differences between visual organs of different animals . However, it is also known that eye development in different animal phyla shares a common genetic network initiated by Pax6 gene orthologs . Similarities in the gene regulatory network that controls eye development, further strengthen the idea that phylogenetically diverse eye types may share a conserved eye developmental program [3–5]. Insects are among the largest and most diverse animal groups. Therefore, decoding eye development in insects offer great opportunity to unravel developmental insights that lead to the emergence of evolutionary complexity.
In most insects, compound eyes represent the prominent visual organs that are responsible for providing major share of the visual information. In the fruit fly Drosophila melanogaster , each compound eye consists of approximately 850 ommatidia and each ommatidium houses eight PRs: six outer PRs and two inner PRs . Additionally, winged insects (such as Drosophila ) possess ocelli that are comparatively simple photosensory organs embedded in the dorsal head cuticle . In Drosophila , a triplet of ocelli (one medial and two lateral) is arranged in a triangular shape between the two compound eyes and the dorsal vertex of the head . It is believed that insect compound eye- and ocellus-like precursor structures have segregated from an ancestral eye over 500 million years ago [9,10]. The evolution of new opsin genes by gene duplication enabled these visual organs to perform different functions that require distinct spectral sensitivities . In Drosophila , phylogenetic analysis supports that Rhodopsin1 (Rh1), Rh2 and Rh6 have originated from a common ancestral gene . A first gene duplication may have separated Rh6 from Rh1/Rh2 and a second gene duplication may have separated the closely related Rh1 and Rh2 . The green-sensitive Rh6 is expressed in the inner PRs of the compound eye and is critically involved in color perception [13–15]. The blue-green sensitive Rh1 is expressed in the outer PRs of the compound eye and is mainly associated with motion detection [16–18]. Conversely, the violet-sensitive Rh2 is expressed in PRs of the ocelli [19–21] and is proposed to be involved in horizon sensing and flight stabilization [22,23]. While it is known that rh1, rh2 and rh6 are genetically linked , it is still unknown how they are differentially expressed in different PRs in Drosophila.
Here we show that the homeodomain transcription factor Homothorax (Hth) regulates Rh2 expression in the ocelli. We demonstrate that Hth is expressed in ocellar PRs and controls a binary rhodopsin switch by promoting Rh2 expression and repressing Rh1 expression in ocelli. We also demonstrate that misexpression of Hth forces outer PR of the retina to induce Rh2 expression and clonal expression in the retina suggest that this process is cell autonomous. Furthermore, genetic and molecular analysis of rh1 and rh2 shows that the rhodopsin switch in ocelli is transcriptionally controlled by Hth and that it may act directly through rh1 and rh2 promoter sequences. Finally, we argue that while Hth maintains Rh3 fate in the DRA , it initiates Rh2 fate in the ocelli. The results presented here greatly adds to our understanding of how genetically linked opsins are spatiotemporally controlled to provide distinct spectral sensitivity to different visual organs.
It is believed that the hexapod ancestor of extant insects only had a single visual organ (also referred as ancestral eye) and that other visual organs (such as compound eye and ocelli) evolved as a result of a morphological bifurcation event over 500 million years ago [10,11] (Fig 1A–1C). It is hypothesised that different functions of compound eye and ocelli in conjunction with gene duplication events of opsins led to the emergence of an ocelli specific opsin gene . In most insects, opsins are categorised based on their distinct spectral sensitivity [26–28]. In the Drosophila compound eye, brightness detection is achieved by inputs from six outer PRs (R1-R6) that express the blue-sensitive Rhodopsin 1 (Rh1). Drosophila uses outer PRs mainly for motion detection and dim light vision [16–18]. Color vision requires two inner PRs that encode either one of the UV-sensitive Rh3 or Rh4 in R7 PRs, and either the blue-sensitive Rh5 or the green-sensitive Rh6 in R8 PRs [29,30]. In most insects, ocelli express an opsin, which is different from those present in the compound eye [11,21,31]. In Drosophila , PRs of the ocelli express the violet-sensitive Rh2 (Fig 1B) [19–21]. The different spectral sensitivity of ocelli is also reflected by their involvement in performing different functions than the compound eye and they are believed to detect horizon, control head orientation and stabilize flight posture while flying . To get an insight into rhodopsin gene duplications in Drosophila, we compared the coding sequences of rh1 to rh6 by generating a phylogenetic tree (by using MUSCLE online tool; phylogeny.fr) . The phylogenetic tree made by the maximum likelihood method showed two clades with branching support value of 1 each (Fig 1D). Clade I showed a tandem gene duplication that separated Rh5 from the closely related Rh3/Rh4 (Fig 1D). Clade II showed a first tandem gene duplication that separated Rh6 from Rh1/Rh2 (Fig 1D). Rh6 subsequently got expressed in the inner PRs of the compound eye (Fig 1J). A further gene duplication led to the separation of closely related Rh1 and Rh2 (Fig 1D). While Rh1 subsequently got expressed in the outer PRs of the compound eye (Fig 1H), Rh2 got exclusively expressed in the ocelli (Fig 1F). To further analyse the evolutionary origin of clade II opsin gene duplications, we generated a phylogenetic tree by using amino acid sequences of Drosophila Rh1, Rh2, Rh6 and their putative orthologs from other dipteran species (Ceratitis capitata or med fly, Musca domestica or house fly, Glossina palpalis or tsetse fly, Lucilia cuprina or Australian sheep blow fly and Aedes aegypti/Anopheles gambiae or mosquitoes; sequences collected from Feuda lab on bitbucket (https://bitbucket.org/Feuda-lab/opsin_diptera/src/master/) . The resulting phylogenetic alignment of Rh1, Rh2 and Rh6 suggests that Rh6 is ancestral to all dipteran species including mosquitoes. A common Rh1/Rh2 ortholog originated from a duplication of Rh6 in the lineage leading to the higher dipterans and a second gene duplication separated Rh1 from Rh2. Further gene duplications of Rh6 occurred in the mosquitoes and of Rh1 in the house fly Musca domestica (S1 Fig).
In the compound eye, the homeodomain transcription factor Homothorax (Hth) is expressed in the dorsal rim area (DRA) of inner PRs, where it has been shown to be both necessary and sufficient to regulate Rh3 expression and thereby critically contribute to the polarized-light sensing system [25,34]. We found that Hth is also expressed in all PRs of the ocelli (Fig 2A). Since ocellar PRs express Rh2, we first analysed if Hth is involved in regulating this expression. We performed knockdown of hth using the pan-photoreceptor driver lGMR -Gal4 and observed a complete loss of Rh2 expression in ocellar PRs (Fig 2B). Additionally, we also observed that in hth knockdown, the loss of Rh2 expression in the ocelli is compensated by the gain of Rh1 (Fig 2B), a Rhodopsin that is normally expressed in outer PRs of the compound eye. Thus, Hth regulates a binary switch of Rhodopsin in the ocelli where it promotes Rh2 expression and represses Rh1 expression.
Hth often acts together with Extradenticle (Exd) [35,36]. Therefore, to test if Exd is also involved in regulation of the binary Rhodopsin switch in the ocelli, we performed knockdown of exd with lGMR- Gal4 and found that it also resulted in the loss of Rh2 expression and gain of Rh1 expression in the ocellar PRs (Fig 2C).
In the compound eye, a feedback mechanism has been described that allows Rh6 to transcriptionally repress Rh5 in yellow R8 PRs . Since in hth knockdown, ectopic Rh1 expression in the ocelli replaces Rh2, we speculated that Hth may repress Rh1 in order to allow Rh2 expression. Knockdown of hth in turn would remove this repression and as a result Rh1 would get activated and repress Rh2 in the ocelli. To check this hypothesis, we misexpressed Rh1 in ocelli by using lGMR -Gal4 and found that ectopic expression of Rh1 was not sufficient to repress Rh2 and in this case ocelli co-expressed both Rh1 and Rh2 (S2A Fig). Thus, loss of Rh2 in hth knockdown seems to be independent of Rh1 repression.
To further test whether switching Rhodopsin expression might also alter PR identity based on other molecular markers of outer PRs in the retina, we monitored Seven-up (svp) and BarH1 expression in the ocelli in hth knockdown. Svp is a steroid hormone receptor and is expressed in R3/R4 and R1/R6 pairs whereas BarH1 is a homeobox transcription factor expressed in the R1/R6 pair of the developing compound eye [38,39]. We found that neither Svp nor BarH1 are expressed in wildtype, nor in hth knockdown ocelli (S3A, S3B, S3C and S3D Fig). Therefore, although ocellar PRs gained Rh1 and lost Rh2 expression when hth was knocked down, they do not seem to undergo an identity change towards retinal outer PRs.
We next investigated whether regulation of this Rhodopsin switch by Hth occurs at the transcriptional level by using ocelli specific rh2-lacZ  and outer PR specific rh1-lacZ  reporter lines. In wildtype control animals, β-gal expression was specifically observed in ocelli in case of rh2-lacZ (Fig 2D) whereas no expression was observed in ocelli in the case of rh1-lacZ (Fig 2G). In hth knockdown background, we found that β-gal expression from rh2-lacZ is completely abolished from the ocelli (Fig 2E), whereas ectopic β-gal expression from rh1-lacZ is now seen in ocellar PRs (Fig 2H). Therefore, Hth is indeed involved in regulating a binary Rhodopsin switch in the ocelli by promoting transcription of rh2 and repressing transcription of rh1.
Since Hth and Exd act together, we next investigated whether Exd is also involved in regulating rh1 and rh2 transcriptionally. We indeed found that β-gal expression from rh2-lacZ was lost (Fig 2F) and ectopic β-gal expression from rh1-lacZ was observed in exd knockdown in ocellar PRs (Fig 2I).
We have previously shown that the homeodomain transcription factor Hazy (Flybase: Pph13 for PvuII-PstI homology 13) controls expression of Rh2 in the ocelli [40,41]. We therefore next asked if Hazy and Hth act jointly to regulate the Rhodopsin switch in ocelli. However, we find that in hazy-/- null mutant flies, absence of Rh2 expression is not accompanied with the ectopic expression of Rh1 in ocellar PRs (S2B Fig). Moreover, we found that Hth is still expressed in the ocelli of hazy-/- mutant flies (S2C Fig). Also, the expression of Hazy remained unchanged in ocellar PRs when knocking down hth (S2D Fig). Thus, the Hth-dependent Rhodopsin switch in the ocelli does not depend on Hazy.
Since Hth encodes a homeodomain transcription factor we next investigated if Hth may act by directly regulating the rhodopsin promotors. Minimal promoter sequences for rh1 and rh2 have been identified previously [39,20]. Like all other Drosophila rhodopsin promoters, they are rather short (300–400 bp) and contain a Rhodopsin-Conserved-Sequence-I (RCSI) element that provides binding sites for Pax6 orthologs and other factors that promote photoreceptor-specific expression . We aligned the minimal promoter sequences of twelve different Drosophila species (S4 and S5 Figs) and found a high level of conservation within the twelve rh1 promoter sequences, while the rh2 promoters were more divergent. A direct alignment of the two promoters was not possible since they have no sequence similarities apart from the RCSI site. A portion of the rh2 minimal promoter sequence overlaps with the coding sequence of the neighbouring gene CG14297 (S5 Fig). We identified three potential Hth binding sites in the minimal promoter region of rh2 (-293/+55) : one within the coding sequence of CG14297, one directly following its Stop codon and one within its 3’UTR. While the sites in the coding sequence and in the 3’UTR are conserved, the site at the Stop codon is only present in the four closest relatives of Drosophila melanogaster. The rh1 minimal promoter region (-247/+73)  contains two potential Hth binding sites upstream of the RCSI (S4 Fig). To test if the Hth binding sites in the rh1 and rh2 promoter regions may be involved in the Rhodopsin switch in ocelli, we created flies containing transgenes with the minimal promoter regions of rh1 or rh2 driving GFP (rh1-GFP and rh2-GFP ). Next, in order to abolish Hth binding, we introduced point mutations in the Hth binding regions of the rh1 and rh2 promoters and created rh1(hth mut)-GFP and rh2(hth mut)-GFP transgenic flies (Fig 3A–3D) (See Materials and Methods for details). In support with the previous observations, we find rh2-GFP expression in the ocelli (Fig 3B) whereas rh1-GFP is not expressed in ocellar PRs (Fig 3E). However, by mutating the Hth binding sites in the promoter region of rh2 (Rh2 (hth mut)-GFP ), we observed a loss of GFP expression in the ocelli (Fig 3C). Conversely, we found that deleting the Hth binding sites in the promoter region of rh1 (rh1 (hth mut)-GFP) leads to ectopic GFP expression in the ocelli (Fig 3F).
In the dorsal rim area of the retina, Hth is required in inner PRs to promote Rh3 expression and ectopic expression of Hth under the control of lGMR -Gal4 was sufficient to block the expression of inner PR Rhodopsins (Rh4, Rh5 and Rh6) and to induce Rh3 expression in all inner PRs. However, in outer PRs, expression of Rh1 was not affected suggesting that only inner PRs were responsive to Hth misexpression . We next investigated if misexpression of Hth in the retina was also sufficient to induce Rh2 expression. In the wildtype retina, Rh1 was uniquely expressed in outer PRs whereas Rh2 was absent (Fig 4A, 4A’ and 4A”). When Hth was misexpressed in the retina under the control of lGMR- Gal4, we observed that Rh2 was now ectopically expressed in all outer PRs. However, we found that Rh1 expression was unaffected and still being expressed in all outer PRs (Fig 4B, 4B’ and 4B”). We also found that ectopic Rh2 expression was only limited to the outer PRs (and not inner PRs) suggesting that only Rh1 expressing PRs were competent to induce Rh2 expression in the retina, while inner PRs were not.
To investigate if this process is cell autonomous, we generated outer PR versus inner PR Hth gain-of-function clones in the retina using the “flip-out” technique  (Materials and Methods for details). Flip-out clones were marked by presence of GFP expression. We found that when clones were generated in the inner PRs, Rh2 expression was not induced (Fig 5A, 5A’ and 5A”). However, when clones were generated in the outer PRs, Rh2 expression was ectopically induced in all GFP expressing clones (Fig 5B, 5B’ and 5B”). Thus, ectopic Rh2 expression in Rh1 expressing PRs of the retina is a cell-autonomous process. Expression of Hth in inner PR clones did not have any effect on Rh1 expression in the neighbouring outer PRs (Fig 5C, 5C’ and 5C”) and Rh1 expression was maintained when clones were generated in the outer PRs (Fig 5D, 5D’ and 5D”).
To understand if Hth cooperates with any of its known interactors to regulate Rh1 and Rh2 expression in the ocelli, we performed a knockdown mini-screen of previously identified Hth interactors (mentioned in Flybase; S6 Fig) by lGMR-Gal4 and assayed for Rh1 and Rh2 expression in ocellar PRs. In agreement with previous findings, Rh2 expression was lost from the ocelli in calmodulin-binding transcriptional activator (camta), longitudenals lacking (lola) and defective proventriculus (dve) knockdown . However, we did not observe a gain of Rh1 in any of these knockdowns (Fig 6B, 6C and 6D). Surprisingly, in the mini-screen, we found that knockdown of the scaffolding protein Scribble (Scrib) [44,45] and the Ets domain transcription factor Ets65A [46,47] resulted in a gain of Rh1 expression in ocellar PRs without changing Rh2 expression (Fig 6E and 6F).
Taken together, we have shown that Hth regulates a binary Rhodopsin switch in ocelli by promoting Rh2 expression at the cost of Rh1. We found that regulation of Rh1 and Rh2 is transcriptionally controlled by Hth acting together with its binding partner Exd and it is independent of Hazy (Fig 7). We also found that Hth regulates the Rhodopsin switch by operating through the promoters of rh1 and rh2. We further demonstrated that misexpression of Hth in the retina modifies outer PRs to gain Rh2 expression resulting in co-expression of Rh1 and Rh2 in outer PRs and we show that this process is cell-autonomous. Finally, by knockdown mini-screen, we identified Scrib and Ets65A as potential repressors of Rh1 expression in the ocelli.
In insects, ocelli represent a fundamentally simpler visual organ whose spectral sensitivity is different from the compound eye. The unique spectral sensitivity of ocellar PRs in Drosophila is provided by the violet-sensing Rh2 . The presence of this particular Rhodopsin exclusively in ocelli and not in the retina may explain why ocelli perform different functions than the compound eye. We have characterized the role of Hth during terminal differentiation of ocellar PRs and showed that Hth acts together with Exd and regulates a binary Rhodopsin switch in ocelli that promotes Rh2 expression and represses Rh1 expression. Hth is known to be expressed in the ocellar primordium of the early third instar larval (L3) eye antennal imaginal disc but gets downregulated later at mid- to late-L3 . However, it is unknown how expression of Hth is re-induced in ocellar PRs. One possible hypothesis could be that a temporal change during metamorphosis such as a pulse of ecdysone hormone with additional signals induces Hth expression in ocelli. In our study, we show that Hth is expressed in all mature and terminally differentiated PRs of the ocelli.
Hth performs similar functions in PRs of the ocelli and the DRA of the retina: in ocelli, it induces Rh2 expression by repressing Rh1 whereas in the DRA of the retina it induces Rh3 by repressing inner PR Rhodopsins . Loss of hth transforms the DRA into odd-coupled ommatidia where Rh3 is expressed in R7 and Rh6 in R8 suggesting that the Rh3/Rh6 pair represents the default state of Rhodopsins in inner PRs of the retina . In ocelli, we find ectopic expression of Rh1 by loss of hth suggesting that Rh1 may be the default state in ocellar PRs. Loss of Hth leads to ectopic expression of the R8 specific transcription factor Senseless in the DRA, which is otherwise not expressed in the wildtype DRA suggesting a fate change of the DRA to become odd-coupled ommatidia . Interestingly, Hth loss in ocelli does not induce expression of outer PR-specific transcription factors (such as Svp and BarH1) suggesting that the ocellar PR fate may not have been changed.
The DNA binding property of Hth is required for proper DRA fate in the retina . Similarly, we observed that Hth regulates a binary Rhodopsin switch in ocelli by transcriptionally controlling rh1 and rh2 expression through their upstream DNA sequences. The homeodomain transcription factor Hazy controls Rhodopsin expression in the retina and in ocelli [40,41]. However, epistatic analysis showed that both Hth and Hazy act independently to control Rh2 expression in ocelli. Misexpression of hth in all developing PRs of the retina is sufficient to induce a fate switch where inner PRs are transformed into PRs of DRA . Rh3 is expanded to all inner PRs whereas specific loss of Rh4, Rh5 and Rh6 was seen. However, outer PR fate was not changed suggesting that only those PRs, which were previously committed to become inner PRs during development, were responsive to Hth . We additionally found that outer PRs, too were responsive to hth misexpression since they induce Rh2 expression while maintaining its default fate by expressing Rh1. However, the genetic program activated by Hth misexpression to induce Rh2 expression in outer PRs is still unknown. One hypothesis could be that Hth directly activates Rh2 expression in the ocelli but it requires an additional cofactor to repress Rh1 expression. This cofactor could be present in the ocelli but not in the retina and this would explain why knockdown of Hth activates Rh1 expression in the ocelli but is not sufficient alone to repress Rh1 expression in the retina. We additionally found that knockdown of Scrib and Ets65A induces Rh1 expression in the ocelli. However, further experiments will be required to establish the role of Scrib and Ets65A as potential repressors of Rh1 expression in ocelli versus retina.
Photoreception is achieved by expression of different opsins in conjunction with an array of proteins of the phototransduction cascade. Each PR contains a specific opsin, a light sensing protein that defines the particular wavelength of light to which a given PR will respond. In most insect species, opsins were mainly categorized based on their spectral sensitivity into: 1) UV-sensitive opsins 2) short-wavelength sensitive opsins and 3) long-wavelength sensitive opsins . Presence of all three opsin types in most insect species may imply that the ancient retina was trichromatic . Further phylogenetic analyses support ancestral trichromacy in insects and show that opsin diversity must have derived from a single ancestral opsin as a result of gene duplications . In many insect species, spectral sensitivity of opsin gene clades is represented by more than one paralog such as the opsin clade consisting of three paralogs in Drosophila (Rh1, Rh2 and Rh6), five in the mosquito Anopheles gambiae and two in the honeybee Apis mellifera . It is believed that a first tandem gene duplication in Drosophila separated Rh6 from its sister paralog Rh1/Rh2 and while Rh6 is retained in the inner PRs of compound eye, Rh1/Rh2 was associated with the outer PRs of retina and PRs of ocelli. A second gene duplication event led to the diversification of Rh1 and Rh2 by accumulation of further amino acid changes in their sequence. While Rh1 expression was retained in the outer PRs of the compound eye, Rh2 became associated with the PRs of the ocelli. Our results also suggest that the ability of Hth to bind to both promoters may have been a prerequisite during evolution to differentially regulate the expression of the two paralogs and thus, provide spectral sensitivity of ocelli by promoting Rh2 expression and repressing Rh1 expression. However, it would be interesting to know if a change in the spectral sensitivity of ocelli upon Hth loss would in turn affect their specific functions. Also, if gene duplications of opsins occurred during evolution to allow the diversification of different spectral sensitivities in different visual organs, it would be interesting to know what kind of spectral sensitivity was present in the ancestral visual organ. We conclude that Hth may act as an evolutionary factor required in the ocelli to provide their unique spectral identity by expression of Rh2. To maintain this unique spectral identity, Hth controls a binary Rhodopsin switch to repress outer PR fate in ocelli by repressing Rh1 expression.
We show here that Hth function both as a transcriptional activator of rh2 and as a transcriptional repressor of rh1. Usually, Hth in conjunction with its binding partner Exd acts as a transcriptional activator. Interestingly, by knocking down exd , we also observed a loss of Rh2 and a gain of Rh1 expression suggesting that both factors act together. It was previously shown that binding of Hth to the DNA requires the presence of the co-factor Exd [50,51]. Therefore, likely regulation of Rh1 and Rh2 depend on the presence of both Hth and Exd binding to the regulatory region. Hth has also been shown to function in gene repression . In this case the formation of the repressor complex occurs directly at the regulatory regions of the repressed gene and depends on the proximity of DNA-binding sites for different components of the complex in the regulatory region. The same mechanism could also explain the opposing regulatory effect that Hth has on rh1 in comparison to rh2 in the ocelli. Analysis of transcription profiles of different PR types would provide a list of Hth interactors expressed in the ocelli versus the outer PRs of the retina. Binding analysis of these interactors to the rh1 and rh2 minimal promoter sequences would help to identify the potential repressor complex forming on the rh1 promoter, helping to better understand the regulatory functions of Hth.
Wildtype Canton S flies have been used in this study. Other fly strains used were: UAS-hthRNAi (BL27655 and BL34637; we do see the Rhodopsin switch phenotype in both RNAi lines. However, all the experiments were done in BL27655), UAS-exdRNAi (BL29338 and BL34897; Rhodopsin switch phenotype was observed in both RNAi lines. However, all experiments were done in BL34897). For knockdown mini-screen, we used UAS-RNAi flies from Bloomington’s Drosophila stock center and the stock numbers are mentioned in S6 Fig. Other fly strains are: UAS-rh1 , UAS-hth , rh1(-252/+57)-lacZ , rh2 (-309/+32)-lacZ , rh2 (-293/+55)-GFP , hazy-/- , hsFLP; lGMR<wt<Gal4; UAS-mCD8::GFP (this “flip-out” Gal4 was a gift from Claude Desplan’s lab). The following transgenic lines were made in this study: rh2 (hth mut)-GFP, rh1-GFP and rh1 (hth mut)-GFP. Flies were reared on standard food medium and at 25°C. Knockdown flies were grown at 29°C temperature. hth overexpression clones were generated by FLP/FRT system by using the “flip-out” technique . A cross of hsFLP; lGMR<wt<Gal4; UAS-mCD8::GFP with UAS-hth was initiated and flies were grown at normal 25°C. Temporal gene expression in the retina was initiated by heat shock at 37°C for half an hour at the late pupal stage to induce expression of the Flip recombinase (FLP). FLP recognises FRT sites in the lGMR<wt<Gal4 cassette and it removes the cassette to activate Gal4 that induced overexpression of hth. After heat shock, vials were placed back at 25°C and freshly hatched flies were dissected. Overexpression clones in the retina were marked by GFP expression.
The rh1 minimal promoter (-247 to +73) was PCR amplified from genomic DNA with primers “rh1 enh Kpn fw” (gcggtacCTGGAGACTCAAGAATAATACTCGGCCAG) and “rh1 enh Xba re” (gatctagAGGGTTCCTGGATTCTGAATATTTCACTG) and cloned into pBluescript vector using the KpnI and XbaI sites added to the primers. For cloning of the rh2 minimal promoter see . The same rh1 and rh2 promoter fragments were in vitro synthesized (BioCat) altering the sequences of the potential Hth binding sites. The first Hth site in the rh1 promoter (TGACAT) was changed to TaAgcT creating a HindIII restriction site. The second Hth site in the rh1 promoter (CTGTCG) was changed to CTaaaG. The first Hth site in the rh2 promoter (GGACAG) was changed to GtttAG, the second Hth site in the rh2 promoter (GTGTCA) was changed into agcTgA and the third Hth site (CTGTCC) was changed to CTaaaC. Both versions of the two enhancers were cloned into a GFP reporter plasmid containing eGFP, a miniwhite marker and an attB site kindly provided by Jens Rister. The plasmids were injected into nos-φC31; attP40 flies for integration on the second choromosome using the φC31 site-specific integration system .
Adult ocelli were dissected and stained by a protocol published in  whereas dissection and immunohistochemistry of adult retinas were done according to . After the immunostainings, tissue samples were mounted by using Vectashield H-1000 (Vector laboratories). Primary antibodies and their dilutions were as follows: Rabbit anti-Rh2 1:100 , Rabbit anti-Rh6 1:10,000 , Rabbit anti-Hth 1:500 , Chicken anti-βGal 1:1000 (Abcam), Chicken anti-GFP 1:2000 (Life technologies), Rabbit anti-Hazy 1:500 , Mouse anti-Svp 1:100 , Rat anti-BarH1 1:200 , Mouse anti-Rh1 1:20, Rat anti-Elav 1:20 and Mouse anti-Chp 1:20 (Developmental Studies Hybridoma bank). Phalloidin conjugated with Alexa-568 and Alexa-647 (Sigma-Aldrich, Life Technologies) marks F-actin and were used (1:5000) during incubation with secondary antibodies. The following secondary antibodies were used: Goat anti-rabbit, Goat anti-mouse, Goat anti-rat and Goat anti-chicken that are conjugated with Alexa-488, Alexa-555 and Alexa-647 (Jackson Immunoresearch). All secondary antibodies are used at 1:200 dilution.
Tissue samples were imaged with a Leica TCS SP5 confocal microscope at a resolution of 1024x1024 pixels and optical sections were taken in the range of 1–2 μm depending on the sample size. Images were further processed and analysed in Fiji/ImageJ and Adobe photoshop 2020 software.
We thank C. Desplan, J. Rister, A. Salzberg, T. Cook, D. Vasiliauskas, Developmental Studies Hybridoma Bank (DSHB) and Bloomington Stock Center for providing flies and antibodies. We thank UniFr bioimage team for maintaining imaging facility and all members of Egger and Sprecher lab for fruitful discussions.