Differential regulation of eye specification in Drosophila by Polycomb Group epigenetic repressors

ABSTRACT

During metazoan development, chromatin plasticity and genetic regulation are tightly connected to establish proper tissue fate and patterning. Drosophila imaginal discs are excellent models to study these processes as both genetic and mechanical injury can redirect their fate during regeneration. Reducing expression of Polycomb (Pc) results in the ectopic activation of the wing selector gene vestigial (vg) which in turn interacts with its DNA-binding partner Scalloped (Sd) – this forces the eye to transform into a wing. Reductions of other Polycomb Group members alone do not phenocopy this transformation. However, knocking down Sex comb on midleg (Scm) or Scm-related gene containing four mbt domains (Sfmbt) alongside the Pax6 gene twin of eyeless (toy) enables the eye-to-wing transformation. Using high throughput sequencing we show that toy-Sfmbt and toy-Scm knockdowns alter expression of both wing selector and Hox genes. These findings provide new insights into how the fate of the eye is specified.

Highlighted Article: Distinct gene sets are differentially expressed in response to the knockdown of Polycomb Group (PcG) members, suggesting that multiple avenues may exist for the eye to be transformed into a wing in Drosophila.

INTRODUCTION

A crucial decision during metazoan development is the acquisition of tissue identity. Failure to correctly specify tissue fate can lead to the derailment of development and the onset of congenital disorders. As such, elucidating the molecular and developmental mechanisms underlying organogenesis is essential for understanding development and disease. The imaginal discs of the fruit fly, Drosophila melanogaster, are popular model systems for studying tissue growth, specification, pattern formation and morphogenesis. These epithelial sacs originate during embryogenesis (Auerbach, 1935; Kaliss, 1939) and give rise to most of the external structures of the adult fly including the head, thorax, legs, wings, halteres and genitalia (Geigy, 1931; Howland and Child, 1935). Because both genetic and mechanical injuries can redirect the fate of the imaginal discs (Weasner and Kumar, 2022), these tissues have become excellent model systems for identifying signal transduction pathways, transcription factors and epigenetic regulators that guide developing tissues through a series of decision points as they make their way to adopting their final fate.

Studies in which cells from developing tissues were disassociated, transplanted into foreign hosts and allowed to develop indicate that imaginal disc fate is determined during embryogenesis (Chan and Gehring, 1971; Gehring, 1970; Hadorn, 1968; Hadorn and Muller, 1974; Schubiger et al., 1969). Similar timings are also associated with the determination of tissue identity in vertebrates (Holtfreter, 1943; Moscona, 1957; Moscona and Moscona, 1952; Trinkaus and Gross, 1961; Weiss and Taylor, 1960). Over the course of three larval instar stages, the imaginal discs grow in size, are regionally specified and undergo pattern formation. While many loss-of-function mutants result in the loss, reduced size, or structural distortion of imaginal disc derivatives, an array of genetic manipulations and mechanical injuries can also force an imaginal disc to undergo transdetermination, the process by which one disc loses its preprogrammed fate and adopts that of another. Each disc has a unique capacity to undergo transdetermination (Hadorn, 1968, 1978; Weasner and Kumar, 2022).

The phenomenon of transdetermination was discovered during studies aimed at understanding how and when the fates of imaginal discs are determined. In short, when whole imaginal discs are transplanted into larval hosts, they will autonomously develop, undergo metamorphosis and produce adult derivatives that can be surgically isolated from the adult hosts (Beadle and Ephrussi, 1935). The resulting adult derivatives are appropriately matched to the transplanted disc (i.e. a wing disc will produce an adult wing, hinge and notum). A similar phenomenon is seen when fragments of imaginal discs are transplanted (Hadorn, 1966, 1967). However, in a small number of examples the transplanted fragment will produce an inappropriate adult derivative. The term transdetermination was coined to describe these changes in fate (Hadorn, 1968). Some discs, such as the antenna, can take on the identity of multiple other discs, whereas others like the eye disc have a much more limited capacity to have its fate altered (Hadorn, 1978; Weasner and Kumar, 2022). Here, we focus on the development of the eye field and its sole propensity to transdetermine into a wing.

An understanding of the molecular mechanisms by which the eye primordium rejects the choice to develop into a wing during normal development is beginning to emerge. This is due in large part to a series of genetic manipulations that mimic the disc fragmentation studies and result in the eye-to-wing transformation. These include (1) double mutants of eyes reduced (eyr) and eyeless (ey) (Edwards and Gardner, 1966), (2) double mutants of opthalmoptera (opht) and ey (Goldschmidt and Lederman-Klein, 1958; Postlethwait, 1974), (3) double mutants of lobiod (ld) and opht (Kobel, 1968; Ouweneel, 1969, 1970), (4) simultaneous expression of the Hox gene Antennapedia (Antp) and activation of the Notch (N) pathway (Kurata et al., 2000), (5) combining the loss of twin of eyeless (toy) with ectopic Antp expression (Gehring et al., 2009), (6) forced expression of the histone modifying factor winged eye (wge) (Katsuyama et al., 2005; Masuko et al., 2018), (7) reduced expression of the Polycomb (Pc) epigenetic repressor (Brown et al., 2023; Zhu et al., 2018), and (8) forced expression of vestigial (vg) – a known wing selector gene (Kim et al., 1996; Simmonds et al., 1998). The roles that Pc and vg play in the eye-to-wing transformation are intertwined with each other, as reductions in Pc expression lead to the upregulation of vg within the eye primordium (Brown et al., 2023; Zhu et al., 2018). This regulation is likely to be direct, as two functional Polycomb Response Elements (PREs) are present within the vg locus (Ahmad and Spens, 2019; Herzog et al., 2014; Okulski et al., 2011; Ringrose et al., 2003). Our current model is that Pc epigenetically silences the vg locus within the eye during normal development thereby preventing it from adopting a wing fate (Brown et al., 2023). At the same time, the retinal determination (RD) network, led by the Pax6 transcription factors Ey and Toy, independently pushes the eye primordium towards a retinal fate (Kumar, 2010). Together, these two mechanisms specify the fate of the compound eye.

In this manuscript we continue our investigation into the role that Polycomb Group (PcG) proteins play in the eye-versus-wing decision by addressing two observations from an earlier study (Zhu et al., 2018). The first observation is that, unlike Pc, knocking down expression of other PcG members individually does not induce an eye-to-wing transformation. This is unexpected, as all PcG members are thought to work together to remodel chromatin and silence gene expression. Second, the eye-to-wing transformation can be induced if select PcG members are simultaneously knocked down with either ey or toy. This is reminiscent of the requirement for ey expression to be lost simultaneously with that of either eyr or opht in order for the fate of the eye to be redirected towards a wing (Edwards and Gardner, 1966; Goldschmidt and Lederman-Klein, 1958; Postlethwait, 1974). As both Ey and Toy are both master regulators for eye specification and paralogs of each other (Czerny et al., 1999; Halder et al., 1995), our findings suggest that their loss enhances developmental plasticity of the disc, which in turn sensitizes the tissue to undergoing transdetermination.

Here we use RNA-sequencing (RNA-seq) and Cleavage Under Targets and Release Using Nuclease (CUT&RUN) to understand why individual knockdown (KD) of Pc, Scm-related gene containing four mbt domains (Sfmbt) and Sex comb on midleg (Scm) has differential effects on eye fate, investigate the impact that the loss of toy has on the transcriptome and chromatin, and compare the molecular signatures of several genotypes that produce the eye-to-wing transformation (Pc KD, toy-Sfmbt double KD and toy-Scm double KD). Our findings suggest that the eye-to-wing transformation in Pc KD and toy-Sfmbt KD arises through a common mechanism; namely, the strong activation of the vg wing selector gene. In contrast, the toy-Scm double KD discs may use a different mechanism that is partially reliant on the activation of Hox genes. Our experimental data also suggest that Toy plays a role in maintaining a robust pattern of H3K27me3 at the vg locus during normal eye development. Levels of this repressive mark are dampened within the eye-antennal discs whenever toy is knocked down (either by itself or in combination with PcG members). Our findings shed new insights into the roles that Pax6 and PcG epigenetic regulators play in establishing the fate of the eye.

RESULTS

Knocking down select PcG members alongside toy drives an eye-to-wing transformation

Our previous investigations into eye-antennal imaginal disc fate have shown that reducing Pc expression leads the eye to transform into a wing (Zhu et al., 2018), and in this context the reduction of Pc in the developing eye field relieves H3K27me3 repression of the vg locus (Brown et al., 2023). This allows for Vg to be expressed, co-opt its binding partner Scalloped (Sd), and force the eye to adopt a wing fate (Brown et al., 2023). Interestingly, although knocking down any of the remaining PcG members fails to drive this transformation, the situation changes when these members are knocked down alongside toy (Zhu et al., 2018). For example, the combined loss of toy with Scm-related gene containing four mbt domains (Sfmbt) (Klymenko et al., 2006; Usui and Simpson, 2000) leads to a drastic eye-to-wing transformation that rivals that of the Pc transformation (Zhu et al., 2018). Here, we ask if the eye-to-wing transformation occurs via the same downstream mechanism and investigate if and how knocking down toy increases the developmental plasticity of the eye field.

To address these questions, we first combined the DE-GAL4 driver (Morrison and Halder, 2010) with UAS based RNA interference (RNAi) lines that target 18 of the 20 PcG members (Fig. S1) and assayed the effects of losing individual PcG members on imaginal discs and adult heads. The DE-GAL4 driver is ideally suited for our studies because, unlike other GAL4 drivers that are expressed broadly across the developing eye (i.e. ey-GAL4, tio-GAL4, eya-GAL4, GMR-GAL4), DE-GAL4 is expressed in a limited spatial domain. Starting at the second larval instar stage, the DE-GAL4 driver is expressed exclusively within the dorsal portion of the eye (Fig. S2). This allowed us to compare the experimental tissue (dorsal) to the control tissue (ventral) simultaneously within the same imaginal disc.

We looked for ectopic expression of the Antp Hox gene and/or a significant morphological change in the structure of the eye field, as these are two hallmark features of PcKD discs (Zhu et al., 2018). The DE-GAL4 driver itself had no effect on Antp expression nor on the morphology of the developing eye-antennal disc (Fig. 1A). Ectopic Antp expression was detected when we knocked down five other PcG members: pleiohomeotic (pho), Calypso, Sex combs extra (Sce), Sfmbt and Scm, even though the eye did not transform into a wing in these instances (Fig. 1B-F). This confirmed our earlier conclusion that, although Antp may play a role in the eye-to-wing transformation, it does not drive this change in fate (Brown et al., 2023). phoKD, CalypsoKD and SfmbtKD discs appeared to be morphologically wild type and ectopic Antp expression was restricted to the peripodial epithelium of these discs (Fig. 1B,C,E). The adult heads of phoKD and SfmbtKD animals also appeared to be wild type and, even though CalypsoKD animals had lost some head capsule, their eyes, ocelli, and antennae appeared normal (Table S1). SceKD discs activated Antp expression specifically within the antennal field and this resulted in its failure to be patterned properly (Fig. 1D, asterisk; Table S1). Finally, ScmKD discs activated Antp in both the peripodial epithelium and disc proper of the entire eye-antennal disc. This led to a slight increase in growth of the dorsal portion of the eye field (Fig. 1F). Knocking down Scm was lethal at the late pupal phase and pharate animals most often failed to properly evolute their heads. Those that did showed a potential wing-like outgrowth from the dorsal eye (Table S1). Overall these results showed that the KD of Pc remained the only example in which the loss of a PcG member, on its own, reproducibly drove the eye to transform into a wing (Fig. 1G, arrow). We have noticed that the development of ectopic wings is limited to the dorsal-anterior quadrant of the eye field when PcG proteins are knocked down (this study; Zhu et al., 2018; Brown et al., 2023). This is true even when more broadly expressed drivers such as ey-GAL4 are used. In contrast, wing tissue arises from a larger spatial domain when vg is ectopically expressed. The Vg-Sd complex induces ectopic wing tissue formation mainly within the dorsal-anterior and ventral-anterior quadrants, with very rare examples of wing development emanating from other regions of the eye field (Brown et al., 2023). Gerold Schubiger proposed that the Wingless and Decapentaplegic signaling pathways are required for transdetermination in leg imaginal discs (Maves and Schubiger, 1995, 1998). Within the eye-antennal disc these two pathways are expressed along both the dorsal and ventral margins of the disc. Their expression patterns do not directly correspond to the location from where ectopic wing tissue emanates. As such, it is not clear why the formation of ectopic wings is relegated to certain ‘hot spots’ or why the KD of PcG members induces wing formation in a more limited spatial zone than when vg is ectopically expressed. We further note that ectopic eyes also develop at particular positions within the eye-antennal disc and that one of the hot spots is close to the vicinity of where the eye transforms into a wing (Salzer and Kumar, 2010).

Fig. 1.

The knockdown of Sfmbt or Scm in addition to toy transforms the eye-antennal disc. (A-G) Third-instar eye-antennal imaginal discs knocking down a PcG member with DE-GAL4; stained for Antp. (A′-G′) Third-instar eye-antennal imaginal discs knocking down a PcG member and toy with DE-GAL4; stained for anti-Antp (green), anti-Elav (blue) and Phalloidin (red). Wild type [DE-GAL4 (A) or DE-GAL4, toy RNAi (A′)] eye-antennal disc from a third-instar wandering larvae. Discs lacking Pho (B), Calypso (C), Sfmbt (E) or Scm (F) alone ectopically express Antp, but otherwise maintain mostly wild-type morphology. The KD of Sce (D) reduces the eye and disrupts the antennal field (asterisk). Knocking down Pc with DE-GAL4 (G, arrow) grants an eye-to-wing transformation similar to that seen with ey-GAL4 (Brown et al., 2023). When combined with toy RNAi, knocking down Pho (B′) and Calypso (C′) leads to an outgrowth in the dorsal field (hatched arrows), while Sce (D′) shows similar eye field disruption to DE>Sce RNAi alone (D). Knocking down toy and Sfmbt (E′) or Scm (F′) gives an exaggerated eye-to-wing transformation (arrows) like that seen in toy-Pc KD discs (G′). (H) Simplified model of the Pc eye-to-wing transformation (left) proposed in Brown et al. (2023), where the presence of Vg and Sd drives the wing transformation (brown) in the dorsal eye field, while continued presence of Pax6 maintains some eye fate (green). The combined loss of toy and Sfmbt, Scm or Pc (right) leads to an ‘enhanced’ eye-to-wing transformation more drastic than that of ey>Pc RNAi or DE>Pc RNAi alone. Please note that we have idealized the drawing of the ectopic wing for simplicity – see the experimental images in the remaining panels. For all phenotypic scoring, n=30. Penetrance and severity of each phenotype are listed in Table S1. Scale bar: 50 μm.

As knocking down toy in conjunction with Sfmbt (toy-SfmbtKD) pushes the eye to adopt a wing fate (Zhu et al., 2018), we examined single toyKD discs to see if the fate of the eye disc is affected. Reducing toy expression, on its own, had minimal effects on the structure of the eye (Fig. 1A′) with only a slight reduction in the number of unit eyes being recorded. However, the loss of Toy does inhibit the formation of the ocelli and inter-ocellar bristles (Zhu et al., 2017; Baker et al., 2018). In contrast, there are significant developmental consequences to the eye-antennal disc when the above six PcG members (pho, Calypso, Sce, Sfmbt, Scm, Pc) were knocked down in conjunction with toy (Fig. 1B′-G′). In each case there appeared to be qualitatively more ectopic Antp protein and increased disc growth. For toy-phoKD and toy-CalypsoKD discs, the increased growth was restricted primarily to the dorsal eye field, while the remainder of the disc appeared to be relatively normal (Fig. 1B′,C′, hatched arrows). toy-SceKD discs contained a possible second antennal segment, while the eye field was diminished in size (Fig. 1D′). And quite surprisingly, toy-SfmbtKD, toy-ScmKD and toy-PcKD discs all showed a massive eye-to-wing transformation, where the transformed wing tissue was no longer constrained to the dorsal eye field, but instead spanned the entire dorsal region of the eye-antennal disc (Fig. 1E′-G′, arrow). It is exceedingly rare for an animal with a full eye-to-wing transformation to survive until adulthood. Many animals die as late larvae or early pupae. Those that survive to the pharate adult stage fail to unfurl their ectopic wings. As the activity of GAL4 is dependent on temperature, we attempted to modulate the level of RNAi expression by raising or lowering the temperature at which KD experiments were conducted. Unfortunately, neither effort yielded adults with fully unfurled ectopic wings. One of the rare exceptions is shown in figure 5H of Zhu et al. (2018).

Since a completely unfurled wing is such a rare occurrence, we sought to confirm that the tissue outgrowths that are seen in the eye-antennal disc are in fact ectopic wing tissue. To do so, we stained control and KD eye-antennal discs with antibodies against the Nubbin (Nub) protein. Nub is expressed through the entire wing pouch (Fig. S3C,J; Ng et al., 1995) and loss-of-function mutants are characterized by the severe reduction or loss of the adult wings (Mickey, 1949; Grell, 1956; Lindsley and Zimm, 1992). In control eye-antennal discs, Nub protein was distributed within the A1 and arista segments of the antenna (Fig. S3A,J). Individual KD of toy, Sfmbt and Scm, which do not exhibit significant tissue outgrowth, also did not show ectopic Nub expression in the eye field (Fig. S3B,E,F,J). This confirms that an eye-to-wing transformation event does not take place in these genetic backgrounds. In contrast, PcKD, toy-PcKD, toy-SfmbtKD and toy-ScmKD discs – which all show dramatic tissue outgrowths – also expressed ectopic Nub in the eye field that resembles a wing pouch (Fig. S3D,G-I,J). This indicates that, although we cannot see perfectly unfurled wings in the adult, the tissue outgrowth in the larval disc is bona fide wing tissue.

In a past study, we demonstrated that the ectopic wing within PcKD discs is patterned correctly across the anterior-posterior (A/P) and dorsal-ventral (D/V) axes (Zhu et al., 2018). Interestingly, these axes are orthogonal to the corresponding axes within the eye disc. This differs from wing duplications, in which the axes of the ectopic wing run in the same direction as the endogenous wing (Zecca et al., 1995). The difference between the two situations may lie in the mechanisms by which the ectopic structure is being generated – i.e., a transformation of one imaginal disc into another versus a duplication of an imaginal disc. To determine if the ectopic wing tissue in toy-SfmbtKD and toy-ScmKD discs are also patterned correctly, we used antibodies against Wingless (Wg) and Cut (Ct) to mark the D/V axis, as well as antibodies against Engrailed (En) and pMAD to mark portions of the A/P axis. The markers were all present within the ectopic wing tissue and, with the exception of En, appeared to recapitulate the normal wing expression pattern (Fig. S4). ct was expressed broadly within the antenna, glial and cone cells of the eye, as well as a stripe along the D/V axis of the wing (Fig. S4A,B, arrow). Within the transformed wing of toy-SfmbtKD and toy-ScmKD discs, Ct appeared in a stripe that marked a new D/V axis (Fig. S4C,D, arrows). pMad protein was distributed along the A/P axis of the eye, antenna and wing discs (Fig. S4E,F, arrow). A stripe of pMad expression was seen within the ectopic wing tissue in both toy-SfmbtKD and toy-ScmKD discs, which indicated the formation of an A/P axis (Fig. S4G,H, arrows). Wg protein was found along the dorsal and ventral margins of the eye field and the dorsal sector of the antennal field (Fig. S4I). It was also expressed in a ring around the wing pouch and in a stripe along the D/V axis of the pouch (Fig. S4J, arrow). Within the transformed tissue of toy-SfmbtKD and toy-ScmKD discs, Wg protein distribution largely recapitulated the normal wing pattern (Fig. S4K,L, arrows). En is normally expressed within the posterior portions domains of the eye and antenna as well as the wing pouch, hinge and notum (Fig. S4M,N, asterisk; Brower, 1986). In the ectopic wing, En protein did not appear to be restricted, but rather appeared to be expressed broadly in the putative ectopic pouch (Fig. S4O,P, asterisks). As such, A/P patterning of the ectopic wing does not completely recapitulate normal wing development.

As we have previously shown that the eye-to-wing transformation in PcKD discs is caused by the ectopic activation of the vg wing selector gene, we wanted to determine whether the other toy-PcG KD transformations also rely on ectopic vg activation or whether sensitized backgrounds with reduced Pax6 (toy) levels engage another molecular mechanism entirely (Fig. 1H). To investigate this question, we used bulk RNA-seq and CUT&RUN to compare Sfmbt, Scm and Pc individual KD discs to toy-PcG double KD discs.

Since the DE-GAL4 driver we used to express RNAi lines was expressed just within the dorsal-anterior region of the mid third larval instar disc (Fig. S2), we were concerned that small but statistically significant changes in gene expression within these cells may be masked by the remaining tissue that did not express the driver. To overcome this potential obstacle, we attempted to use the Isolation of Nuclei Tagged in specific Cell Types (INTACT) system (Deal and Henikoff, 2010; Ma and Weake, 2014) to specifically isolate nuclei from a genetically appropriate cell population for both RNA-seq and CUT&RUN. The version of the technique that we used in our study uses a UAS-GFP-KASH (Klarsicht/Anc-1/Syne homology) (Starr, 2011) transgene to label the outer nuclear envelope in a driver-dependent context. This allows for nuclei to be isolated using a combination of GFP antibodies and bead-based purification technologies. Driving UAS-GFP-KASH in the eye-antennal and wing discs with DE-GAL4 yielded the expected expression patterns when only toy is knocked down (Fig. S5A,B). However, incorporating the additional KD of Sfmbt – which triggers the eye-to-wing transformation – severely disrupted the GFP-KASH spatial expression pattern (Fig. S5C,D, arrows). In control wing discs, the DE-GAL4 driver activated reporter expression only within the hinge, rather than the pouch (Fig. S5B). When the dorsal eye field of the toy-SfmbtKD disc began to transform into a wing, the expression GFP-KASH within the dorsal-anterior quadrant of the eye was abrogated and was activated just within the hinge domain of the ectopic wing. This left the ectopic wing pouch without GFP-KASH expression. It was also possible that non-autonomous signaling from the ventral compartment of the eye field inhibited expression of the DE-GAL4 driver. We thus abandoned using the INTACT system and instead focused on finding the time point at which the eye-to-wing transformation in each genotype reached its zenith in terms of size. At these time points, the largest double KD discs which were likely to have the most complete eye-to-wing transformation would be selected for RNA-seq and CUT&RUN analysis.

To determine the point in development at which each disc is at the apex of its transformation, we examined control, single KD and double KD discs at various stages of third larval instar development (Fig. 2). Wild-type animals, when reared at 25°C, complete larval development and begin pupariation around 120 h after egg lay (AEL). This is also the case for ScmKD and SfmbtKD discs in which ectopic Antp expression had been activated but the eye-to-wing transformation was not induced (Fig. 2B,C, blue box). On the other hand, the onset of pupariation was delayed by 24 h in PcKD, toy-PcKD, toy-ScmKD and toy-SfmbtKD discs where the eye had transformed into a wing (Fig. 2A,A′-C′, red boxes). While PcKD and toy-PcKD discs consistently reached their peak size by 144 h AEL, toy-SfmbtKD and toy-ScmKD discs were on average smaller (Figs 1E′,F′, 2B′,C′). In fact, only a small fraction of toy-SfmbtKD and toy-ScmKD discs reached their full size by 144 h AEL. We attempted to isolate toy-SfmbtKD and toy-ScmKD discs 12 h (156 h AEL) and 24 h (168 h AEL) later, but the vast majority of both types of larvae pupated by 156 h. Since wild-type discs continued to grow and undergo patterning during the first 12 h of pupal development (Wolff and Ready, 1993), we speculated that this may also be the case for toy-SfmbtKD and toy-ScmKD discs as well. In the end, we decided to isolate discs from DE-GAL4 (control), toyKD, SfmbtKD and ScmKD larvae that were 120 h old while isolating discs from PcKD, toy-PcKD, toy-SfmbtKD and toy-ScmKD larvae that were 144 h old. The size of discs containing ectopic wing tissue varied, but the precise staging helped to minimize potential variance. We note here that, as these discs were being dissected at such late time points within the third larval instar, it was likely that we would identify several classes of genes including those that are crucial for both specification and construction of the wing.

Fig. 2.

The eye-to-wing transformation occurs consistently at 144 h AEL. (A-C′) Time course of DE-GAL4>UAS-PcG RNAi (A-C) and DE-toy RNAi>UAS-PcG RNAi (A′-C′) at the end of third-instar development (120 h-144 h AEL); stained with anti-Antp (green), anti-Elav (blue) and Phalloidin (red). When knocked down with DE-GAL4 alone, Pc RNAi animals delay pupation for an additional 24 h and the dorsal eye field transforms by 144 h AEL (A). Development of Scm (B) and Sfmbt (C) KD animals is not delayed, and these animals pupate after 120 h with minimal eye-antennal disc transformation. Upon combined loss of toy with Pc (A′), Scm (B′) or Sfmbt (C′), all animals are developmentally delayed and transform by 144 h AEL. Arrows indicate an eye-to-wing transformation in the dorsal eye field. Boxed eye-antennal discs represent time points of transformed (blue) and non-transformed (red) discs that were selected for RNA-seq. Scale bar: 50 μm.

Bulk RNA-seq reveals few expression changes when toy alone is knocked down

To better understand the molecular mechanisms underlying the three robust examples of eye-to-wing transformation events, we first used bulk RNA-seq to compare the transcriptome profiles of PcKD discs to ScmKD or SfmbtKD discs that are morphologically wild type and to toy-ScmKD or toy-SfmbtKD discs that harbor the eye-to-wing transformation (Table S2). Transcript counts showed that Pc, Sfmbt and Scm were reduced in the respective single and double KDs – this confirmed that the RNAi lines used do indeed knock down expression of their targets (Fig. S6A-C). These counts also indicated that the KD of one PcG member did not affect the expression of other PcG members. This suggests that the phenotypes that we observed are due to the specific KD of toy and/or individual PcG members. Hierarchical clustering analysis shows that the transcriptomes of morphologically ‘wild-type’ eye-antennal discs (orange) discs tended to cluster together and were separated from the eye-antennal discs of those that had been transformed into a wing (purple) (Fig. 3A). Interestingly, there were outliers with each of these two broad categories. One example was that two of three biological samples of SfmbtKD discs (which are morphologically wild type) clustered with the genotypes that had the eye-to-wing transformation instead of the other genotypes that had morphologically normal discs. Another irregularity was that the three biological replicates of each genotype did not always cluster together (see replicates of the DE-GAL4 control discs).

Fig. 3.

The eye-to-wing transformation clusters separately from morphologically wild-type eye-antennal discs. (A) CPM normalized transcript counts of 120 h AEL ‘wild-type’ (orange: DE-GAL4, DE-GAL4>UAS-toy RNAi, DE-GAL4>UAS Sfmbt RNAi and DE-GAL4>UAS-Scm RNAi) and transformed (purple: DE-GAL4>UAS-Pc RNAi, DE-toy>UAS-Pc RNAi, DE-toy>UAS-Sfmbt RNAi and DE-toy>UAS-Scm RNAi) eye-antennal discs. Heatmap depicts hierarchical clustering of the top 100 most variable genes. (B) MA plot of differentially expressed genes between wild-type (DE-GAL4) eye-antennal discs and DE-toy RNAi (toyKD) (n=108) eye-antennal discs. (C) Enriched GO analysis of differentially expressed genes between wild-type (DE-GAL4) eye-antennal discs and toyKD eye-antennal discs.

Several factors may contribute to the irregularities that we observed in the clustering analysis. First, the DE-GAL4 line results from an insertion of GAL4 into the mirror (mirr) locus which results in a loss-of-function mutant. DE-GAL4 homozygotes die during embryogenesis but DE-GAL4/+ heterozygotes (which are used in this study) are viable with no overt visible effects on the structure of the eye (Morrison and Halder, 2010). Second, reducing toy levels with RNAi reduces the number of ommatidia slightly and eliminates the ocelli and inter-ocellar bristles (Zhu et al., 2017; Baker et al., 2018; Teeters et al., 2024). Third, KD of the PcG proteins Sfmbt and Scm on their own, while not inducing an eye-to-wing transformation, did in fact affect the structure and shape of the imaginal disc (Fig. 2A-C). Fourth, discs that contained eye-to-wing transformations were of significantly distinct sizes (Fig. 2A′-C′). These factors are likely to have some effect on the transcriptome and could be the reason why some biological replicates did not cluster appropriately. We attempted to minimize the differences between samples by carefully staging the larvae (Fig. 2) and by selecting discs during the dissection process that were of similar sizes.

In order to evaluate the quality of our RNA-seq datasets we examined the expression of several eye, antenna and wing genes within eye-antenna and wing discs from control and DE-GAL4, UAS-toy RNAi animals (Fig. S7). We examined the expression of several genes that control eye development and were typically expressed either ahead of the morphogenetic furrow [ey and eyegone (eyg)], in both undifferentiated cells and differentiating photoreceptors [eyes absent (eya), sine oculis (so) and dachshund (dac)], or exclusively in postmitotic neurons [embryonic lethal abnormal vision (elav)]. All six of these genes were expressed robustly within the eye-antennal disc and weakly within the wing disc. This is consistent with the known expression patterns of these genes. For the antenna, we examined the expression of the distal antenna (dan) and ct genes. These too were expressed strongly within the eye-antennal disc and weakly within the wing disc. In contrast, the expression of three wing selector genes – vg, nub and apterous (ap) – was elevated in the wing disc and downregulated within the eye-antennal disc.

As DE-GAL4 control, toyKD, ScmKD and SfmbtKD discs are morphologically wild type (Figs 1A,A′ and 2B,C), it is not surprising that the transcriptomes of these control and mutant genotypes clustered together (Fig. 3A). The relatively normal state of these discs was reflected in the small number and the nature of genes that were differentially expressed. This is best illustrated by the toyKD discs in which only 108 transcripts were differentially expressed when compared to DE-GAL4 control discs (Fig. 3B). Most of these differentially expressed genes were unannotated protein-coding (CG) or non-protein-coding (CR) genes. The top Gene Ontology (GO) terms associated with these genes implicated them in sensory perception processes (Fig. 3C). As these genes were not expected to play roles during imaginal disc development it is no surprise that the discs were morphologically normal. Likewise, we did not observe drastic differences between the transcript profiles of PcKD and toy-PcKD discs (Fig. S8A). Despite the increased level of tissue proliferation that occured in toy-PcKD discs, only 69 additional genes were differentially expressed and most of these were annotated as being involved in cuticle development, sensory perception or metabolic processes (Fig. S8A). It suggests that Toy does not play additional roles in transdetermination beyond suppressing the eye-to-wing transformation.

This was not necessarily the case when comparing SfmbtKD or ScmKD non-transformed discs to their respective toy-SfmbtKD and toy-ScmKD counterparts, which is consistent with the eye-to-wing transformation occurring within the double KD discs. When comparing SfmbtKD and toy-SfmbtKD discs to each other, we noted that an additional 209 transcripts were differentially expressed. While the top GO terms associated with these genes implicated them in small molecule biosynthetic and various metabolic processes (Fig. S8B), we noted that the wing selector gene vg was also upregulated in toy-SfmbtKD discs. As vg is responsible for driving the eye-to-wing transformation in PcKD discs (Brown et al., 2023), our identification of vg here could be an indication that the eye-to-wing transformation in both PcKD and toy-Sfmbt KD discs both involve ectopic activation of vg and the formation of the Vg-Sd complex. In contrast, vg was not one of the 144 differentially expressed genes found in toy-ScmKD discs, which are mostly annotated as being involved in protein folding (Fig. S8C). Instead, several Hox genes including Antp, Sex combs reduced (Scr) and Ultrabithorax (Ubx) were differentially expressed. The differences between differentially expressed genes within toy-Sfmbt and toy-Scm discs may provide insight as to why the toy-ScmKD discs clustered separately from the toy-SfmbtKD and toy-PcKD discs (Fig. 3A). It may also suggest that the eye-to-wing transformation in this context occurs via a mechanism that is independent of the Vg-Sd complex.

To better understand the potentially stepwise transcript changes underlying the eye-to-wing transformation, we next made two additional transcriptome comparisons. The first was to compare the profile of DE-GAL4 control discs to those of single PcG KDs (Fig. 4; Table S3). In our past study of PcKD discs we selected transcripts that were upregulated in both wild-type wing and PcKD transformed eye-antennal disc but downregulated in the wild-type eye-antennal disc (Brown et al., 2023). Here, we instead analyzed all 645 genes that are differentially expressed in PcKD discs when compared to DE-GAL4 control discs (Fig. 4A,B). In these discs several Hox genes, including Antp, Ubx, abdominal-A (abd-A) and Abdominal-B (Abd-B), as well as numerous wing selection genes such as vg, nub, ap and wg, were upregulated, while the Hox gene, Deformed (Dfd) was downregulated. Interestingly, despite the repression of eye development in the dorsal region of the disc, the expression of RD network members – including toy – was not significantly downregulated in these discs. This finding suggests that the loss or reduction of RD genes (and by extension eye fate) is not a requirement for the eye to transform into a wing. This is consistent with the ability of PcKD discs to undergo a fate transformation without the need for toy to be simultaneously knocked down.

Fig. 4.

Transformed Pc knockdown discs share gene expression changes with Sfmbt and Scm knockdown discs. (A) Venn diagram showing overlap of differentially expressed genes (Padj<0.05) in PcKD (maroon, n=645), SfmbtKD (green, n=326) and ScmKD (blue, n=762) eye-antennal discs (middle). Up- and downregulated genes in each category are listed in Table S3 (sheet 1). GO analysis for genes shared between Pc, Sfmbt and Scm (right), Pc and Scm (left), or Pc and Sfmbt (top). (B-D) MA plots of differentially expressed genes between wild-type (DE-GAL4) and Pc (B), Sfmbt (C) or Scm (D) KD eye-antennal discs – top differentially expressed genes are labeled.

As the eye field did not transform in either SfmbtKD or ScmKD discs, we expected that fewer genes would be differentially expressed in these discs when compared to DE-GAL4 control discs. This was definitely true for SfmbtKD discs, in which only 326 transcripts (versus 645 in PcKD discs) were differentially expressed between these two genotypes (Fig. 4C). Interestingly, the minimal amount of ectopic Antp expression that we initially observed in these discs (Fig. 1E) was not significant by DESeq2 parameters (Padj<0.05). Instead, the only Hox gene that was differentially expressed was Abd-B. In contrast, ScmKD discs unexpectedly displayed 762 differentially expressed transcripts when compared to DE-GAL4 control discs (Fig. 4D). The loss of Scm within the eye field resulted in the misregulation of several Hox genes including Dfd, Antp, Ubx, abd-A and Abd-B, which was very similar to what has been observed for PcKD discs. Not surprisingly, neither vg nor any other wing selector gene was differentially expressed in either SfmbtKD or ScmKD discs. The lack of ectopic vg expression is the most likely explanation for why the eye failed to transform into a wing in these two PcG member KDs.

The second comparison that we made was between toyKD discs and toy-PcKD, toy-SfmbtKD and toy-ScmKD discs (Fig. 5; Table S3). As the eye transforms into a wing in all three toy-PcG KD discs we expected that the three toy-PcG double KD genotypes would share a larger proportion of differentially expressed genes than the individual KD discs. However, this was not the case (compare Fig. 4A to 5A; Tables S2, S3) and we instead observed that slightly more differentially expressed transcripts were shared between all three individual PcG KD discs than the three toy-PcG KD discs. The transcript profiles of toy-PcKD discs differed from toyKD discs by the differential expression of 412 genes (Fig. 5B). toy-PcKD showed a similar differential regulation of Hox (up: Scr, Antp, Ubx, Abd-B; down: lab, Dfd) and wing (vg, nub, ap) genes. The transformed toy-SfmbtKD and toy-ScmKD discs also showed drastically different expression patterns (314 and 800 genes, respectively) when compared with the single toyKD disc. However, these two groups then appeared to bifurcate, where toy-SfmbtKD discs misregulated the same set of wing genes as toy-PcKD discs, but only upregulated the Antp and Abd-B of Hox genes (Fig. 5C). On the other hand, toy-ScmKD discs misregulated six of the eight Hox genes, but failed to differentially express vg (Fig. 5D). These patterns of gene expression are interesting as they clearly suggest that there may be multiple molecular paths by which the eye-antennal disc can have its fate redirected towards a wing. This model is consistent with the relatively eclectic collection of genetic backgrounds (listed in the introduction) that produce an eye-to-wing fate change.

Fig. 5.

Transformed Pc knockdown discs are more similar to Sfmbt knockdown discs. (A) Venn diagram showing overlap of differentially expressed genes (Padj<0.05) in toy-PcKD (maroon, n=412), toy-SfmbtKD (green, n=251) and toy-ScmKD (blue, n=800) eye-antennal discs (middle). Up- and downregulated genes in each category are listed in Table S3 (sheet 2). GO analysis for genes shared between toy-Pc, toy-Sfmbt and toy-Scm (right), toy-Pc and toy-Scm (left), or toy-Pc and toy-Sfmbt (top). (B-D) MA plots of differentially expressed genes between toy-Pc (B), toy-Sfmbt (C) or toy-Scm (D) double KD discs alone and the toyKD discs. vg is differentially expressed in toy-Pc and toy-Sfmbt KD discs, but not toy-Scm KD discs.

PcKD and toy-SfmbtKD discs share more similarities than toy-ScmKD discs

To determine whether there is more than one path by which the eye can transdetermine into a wing, we compared the transcriptomes of PcKD discs to both Sfmbt and Scm single KD discs and toy-Sfmbt and toy-Scm double KD discs. Differential expression analysis (log2 fold change, Padj<0.05) of these tissues identified 503 genes that were differentially expressed between PcKD and SfmbtKD, 458 genes between PcKD and ScmKD, 252 genes between PcKD and toy-SfmbtKD, and 657 genes between PcKD and toy-ScmKD (Fig. 6A-F). This indicates that of all the tested comparisons, the transcriptomes of toy-SfmbtKD discs may be the most similar to that of PcKD discs, while toy-ScmKD discs may be the most distinct. This correlates with our observations that toy-ScmKD discs cluster separately from the other genotypes that produce eye-to-wing transformations (Fig. 3A).

Fig. 6.

Previously identified eye-to-wing candidates are differentially expressed in toy-Sfmbt and toy-Scm knockdown discs. (A,B) Heatmap of log2 transformed differential transcript expression between PcKD eye-antennal discs and either SfmbtKD and ScmKD eye-antennal discs (A) or toy-SfmbtKD and toy-ScmKD eye-antennal discs (B). Known eye genes, wing genes and Hox genes are listed to the right. (C-F) MA plots of differentially expressed genes between PcKD discs and SfmbtKD (C, n=503), ScmKD (D, n=458), toy-SfmbtKD (E, n=252) or toy-ScmKD (F, n=657) discs. The top ten differentially expressed genes as well as known wing, eye and Hox genes are annotated. (G,H) Venn diagram of known Pc eye-to-wing candidates (maroon, Brown et al., 2023) compared to differentially expressed genes in SfmbtKD (green) and ScmKD (blue) discs (G) or toy-SfmbtKD (green) and toy-ScmKD (blue) discs (H). Overlapping candidates are listed below, with most likely candidates (shared between early, mid and late development) highlighted in red.

In our previous study of the PcKD discs, we identified 95 candidates that were differentially expressed during either early, mid or late third-instar development (Brown et al., 2023). To determine whether any of these candidates play roles in the toy-dependent eye-to-wing transformations, we asked if any of these genes are also differentially expressed in SfmbtKD and ScmKD discs (Fig. 6G) or in toy-SfmbtKD and toy-ScmKD discs (Fig. 6H). Although ten differentially expressed candidate genes are shared between PcKD and either SfmbtKD or ScmKD discs, the only one that was differentially expressed in all three types of discs was Abd-B. Similarly, of the 15 differentially expressed candidate genes that are shared between PcKD and either toy-SfmbtKD or toy-ScmKD discs, only the Hox genes Antp and Abd-B, as well as the wing selector gene ap, were found in all three transformed discs (Fig. 6H). These are unlikely to drive the eye-to-wing transformation, as our previous work indicates that the forced expression of these Hox genes severely abrogates retinal development without inducting the eye-to-wing transformation. Likewise, ectopic expression of ap does not induce the eye-to-wing transformation (Brown et al., 2023). The wing selector gene, vg, that we have shown to drive the Pc-dependent eye-to-wing transformation was only upregulated in toy-SfmbtKD discs (Fig. 6H, red text). This may suggest that in both PcKD and toy-SfmbtKD discs, the ectopic formation of the Vg-Sd complex drives the transformation of the eye into a wing while a distinct mechanism may drive this change in fate in toy-ScmKD discs. Of note, Antp expression was upregulated in toy-ScmKD discs. The combination of Antp expression and toy reduction is known to induce an eye-to-wing transformation (Gehring et al., 2009), thus it is also possible that this genetic combination is the underlying mechanism for the eye-to-wing transformation in toy-ScmKD discs. A second possibility is that the transformation of an eye into a wing may be driven by the combination of knocking down toy and upregulating ap. Ap sits upstream of vg within the wing gene regulatory network and Vg controls only a subset of all Ap functions (Delanoue et al., 2002). It is possible that the simultaneous loss of toy and upregulation of ap could bypass the requirement for the Vg-Sd complex and force the eye to transform into a wing.

Reductions in toy dampen H3K37me3 marks at the vg locus

To gain insight into how Toy mediates the eye-to-wing transformation we used CUT&RUN to analyze H3K27me3 levels in eye-antennal discs of control and transformed larvae (Fig. 7A; Fig. S9). We focused on the vg locus as it is activated in both PcKD and toy-SfmbtKD discs. H3K27me3 heavily decorated the vg locus in DE-GAL4 control discs (Fig. 7A, first row). In PcKD discs this methylation mark was drastically reduced but not completely eliminated (Fig. 7A, third row). As a result, vg expression was ectopically activated within the eye field (Fig. 7B) (Brown et al., 2023; Zhu et al., 2018). Interestingly, H3K27me3 levels were significantly dampened when toy expression was knocked down alone (Fig. 7A, second row). This resulted in a slight increase in vg expression (Fig. 7B,C) but no change in the fate of the eye (Fig. 1A′) (Zhu et al., 2017). Moreover, H3K27me3 marks at the vg locus were depressed even further when toy-PcKD and PcKD discs were compared to each other (Fig. 7A, third and fourth rows). This is consistent with an even higher increase in vg expression (Fig. 7B,C) and a more extensive eye-to-wing transformation in toy-PcKD discs (Figs 1G,G′ and 2A,A′). Finally, while the vg locus within SfmbtKD discs remained heavily decorated with H3K27me3 marks (Fig. 7A, fifth row), those levels dropped significantly in toy-SfmbtKD discs (Fig. 7A, sixth row). In fact, the levels of H3K27me3 in toy-SfmbtKD discs were even lower than those in toyKD discs (Fig. 7A, second and sixth rows). The further depression in H2K27me3 levels appeared to be sufficient to allow for activation of vg expression in toy-SfmbtKD discs (Fig. 7B,C) and the eye-to-wing transformation (Fig. 1E′) (Zhu et al., 2018). Interestingly, the vg locus remained heavily decorated with H3K27me3 in both ScmKD and toy-ScmKD discs (Fig. 7A, seventh and eight rows). This is consistent with RNA-seq data showing minimal to no expression of vg (Fig. 7B). The expression levels of vg in control and KD discs were validated using qRT-PCR (Fig. 7C). As such, we conclude that a potential role for Toy (and possibly Ey) during normal eye development could be to participate with select PcG members in the maintenance of robust levels of H3K27me3 at the vg locus.

Fig. 7.

RNAi perturbation reduces H3K27me3 levels at vg locus. (A) CUT&RUN H3K27me3 peaks at the vg locus in either the ‘wild-type’ (DE-GAL4) eye-antennal disc (blue) or toyKD eye-antennal disc (green) crossed to Pc, Sfmbt or Scm RNAi lines. (B) Bar graph represents transcript counts of three replicates for the DE-GAL4, PcKD, SfmbtKD or ScmKD (blue), and toyKD, toy-PcKD, toy-SfmbtKD or toy-ScmKD eye-antennal discs (green). (C) Bar graph depicting fold changes in vg expression obtained using qRT-PCR on three biological replicates and three technical replicates for the DE-GAL4, PcKD, SfmbtKD or ScmKD (blue), and toyKD, toy-PcKD, toy-SfmbtKD or toy-ScmKD eye-antennal discs (green). Data are mean±s.d.

DISCUSSION

In this manuscript, we have continued our investigation into the role that PcG epigenetic repressors play in regulating the choice between eye and wing fates in Drosophila. Our past studies have shown that, in the absence of Pc expression within the eye field, H3K27me3 histone modifications across the vg locus are abrogated and vg is inappropriately expressed. This results in the ectopic formation of the Vg-Sd complex, which then goes on to reprogram the eye into a wing (Brown et al., 2023; Zhu et al., 2018). In the course of these studies, we made two unexpected observations. First, reducing the expression of other PcG members did not have the same effect on the fate of the eye. Second, the eye could be redirected towards a wing fate if the expression of select PcG members was simultaneously reduced along with that of the Pax6 genes ey and toy. Here, we have focused on the mechanisms by which the eye transforms into a wing when toy levels are simultaneously reduced with either Sfmbt or Scm – two PcG members.

We undertook this investigation because all PcG members are thought to function together to silence target genes (Fig. S1). This is particularly true of Pc, Sfmbt and Scm. Sfmbt and pho encode members of the Pho-repressive complex (PhoRC) (Kassis et al., 2017). This complex binds to its molecular targets by recognizing PRE elements (Brown et al., 1998; Klymenko et al., 2006). PhoRC physically interacts with both PRC1, of which Pc is a member, and PRC2. PhoRC-PRC1 interactions are mediated by Scm, which acts as physical bridge between Sfmbt and Polyhomeotic (Ph), a member of PRC1 (Frey et al., 2016; Grimm et al., 2009; Peterson et al., 1997). PhoRC-PRC2 interactions are mediated by physical contacts between Pho and Enhancer of zeste [E(z)] (Wang et al., 2004). As PhoRC, PRC2 and PRC1 all converge at PRE sites, it is surprising that individual reductions in the expression of all three complex members do not have the same effect on the fate of the eye. Instead, our observations suggest that eye specification is differentially regulated by individual PcG proteins. Our transcriptome analysis of Pc, Sfmbt and Scm single KD discs confirms this hypothesis, as its shows that expression of the vg wing selector gene (confirmed by qRT-PCR) is upregulated only in PcKD discs and not within either SfmbtKD or ScmKD discs.

We also attempted to address the role that Toy plays in maintaining the fate of the eye field. We first asked whether Toy plays a role in regulating vg expression either on its own or in combination with PcG members. Toy appears to function solely as a transcriptional activator (Weasner et al., 2009), which is inconsistent with a potential role in repressing vg expression in the eye disc. Indeed, our RNA-seq data indicate that vg expression is not differentially expressed in toyKD eye-antennal discs. While this data on its own would indicate that Toy does not regulate vg expression, we did find two examples in which the loss of Toy did paradoxically result in an upregulation of vg expression. First, while vg expression is not differentially expressed in SfmbtKD discs, it is ectopically expressed at significantly higher levels in toy-SfmbtKD discs. Second, although vg is differentially expressed in PcKD discs, its expression levels are even higher in toy-PcKD discs. How can the transcriptome results of toyKD discs be reconciled with those of toy-SfmbtKD and toy-PcKD discs? The answer may lie within our CUT&RUN data on H3K27me3 histone marks. As noted above, reducing toy expression in either single or double KD conditions results in a dampening of the H3K27me3 mark at the vg locus. The weakening of H3K27me3 levels in response to knocking down toy may be responsible for the elevation of vg expression levels. We note that, while H3K27me3 levels are strongly reduced in toyKD discs, it is not sufficient, on its own, to significantly alter either vg expression or the fate of the eye. It is possible that the reductions in toy expression reduce H3K27me3 levels to a sensitized threshold that can be further modified by the additional loss of PcG factors. How would Toy influence H3K27me3 levels? One exciting and intriguing possibility that requires further experimentation is that Toy could participate in the recruitment of PcG proteins to PRE elements. If this were to be the case it would imply that Toy may provide some specificity to PcG protein binding within the genome of eye-antennal disc. It would also implicate Toy in the direct repression of wing fate.

Lastly, we addressed the question of whether there are multiple paths for the eye to transform into a wing. As we describe in the introduction to this paper, eight different genetic combinations that induce an eye-to-wing transformation are found in the published literature and we present several additional genetic combinations that do the same. It is unclear if all of these genetic combinations have the same transcriptomic effect. We do not address all known genotypes that induce the transformation of the eye into a wing but instead focus on PcKD, toy-PcKD, toy-SfmbtKD and toy-ScmKD combinations. Our transcriptomic analysis indicates that vg expression is ectopically upregulated in all but toy-ScmKD discs. As we have noted above, this suggests that the eye-to-wing transformation observed in PcKD, toy-PcKD and toy-SfmbtKD discs are likely occurring through the same molecular mechanism – namely through the ectopic formation of the Vg-Sd complex (Fig. 8). In contrast, the eye transforms into a wing in toy-Scm double KDs despite the lack of vg expression. This suggests that that a second molecular pathway, one that is independent of the Vg-Sd complex, may be engaged in the eye primordium when toy and Scm are simultaneously knocked down (Fig. 8). Our transcriptomic analysis of toy-ScmKD discs indicates that numerous Hox genes, as well as the wing selector gene ap, are differentially expressed. While not directly tested here, it is possible that a combination of these factors drives the eye to transform into a wing. The potential existence of multiple molecular paths is consistent with Waddington's epigenetic landscape model (Waddington, 1957) in which an individual tissue can deviate from its original trajectory and embark on several distinct alternate paths as it undergoes transdetermination and/or transdifferentiation.

Fig. 8.

Model figure for Toy and PcG control of eye fate. We have investigated four genetic methods which transform the eye into a wing. (1) individually knocking down Pc (PcKD); (2) knocking down toy and Pc (toy-PcKD); (3) knocking down toy and Sfmbt (toy-SfmbtKD); and (4) knocking down toy and Scm (toy-ScmKD). Based on the bulk RNA-seq data presented here, methods 1-3 appear to rely on ectopically expressing vg to force wing fate. However, method 4 may rely on a vg-independent mechanism.

MATERIALS AND METHODS

Genetics

The following fly stocks were used from the Bloomington Drosophila Stock Center (BDSC; Indiana University, IN, USA): UAS-Pho RNAi, BL42926; UAS-Calypso RNAi, BL56888; UAS-Sce RNAi, BL67924; UAS-Sfmbt RNAi, BL32473 and BL28677; UAS-Scm RNAi, BL55278; UAS-Pc RNAi, BL33964; UAS-toy RNAi, BL33679; UAS-KASH GFP, BL92582; vg¹ BL432; and dorsal eye-GAL4 (De-GAL4; Georg Halder, Katholic University, Leuven, Belgium). Flies were grown on standard cornmeal food and all experiments were conducted at 25°C. For scoring of all eye-antennal discs and adult phenotypes, n=30.

Immunofluorescence and imaging

Drosophila eye-antennal imaginal discs were prepared as previously described (Spratford and Kumar, 2014). The following antibodies were used: rat anti-Elav [Developmental Studies Hybridoma Bank (DSHB), #7E8A10, 1:100]; mouse anti-Antp (DSHB, #8C11, 1:100); mouse anti-Nubbin (DSHB, #2D4, 1:50); mouse anti-Wingless (DHSB, #4D4, 1:100); mouse anti-Engrailed (DSHB, #4D9, 1:100); mouse anti-Cut (DSHB, #2B10, 1:100); rabbit anti-pMad (Cell Signaling Technologies, #41D10, 1:250); FITC-conjugated donkey anti-rat (Jackson ImmunoResearch, #712-095-153, 1:100); Cy3-conjugated donkey anti-mouse (Jackson ImmunoResearch, #715-165-151); Alexa Fluor 647-conjugated Phalloidin (Thermo Fisher Scientific, #A22287, 1:20); and Rhodamine Phalloidin (Thermo Fisher Scientific, #R415, 1:100). Light microscope fluorescent images were taken using a Zeiss Axioplan II compound microscope and processed by Fiji/ImageJ (Schindelin et al., 2012) and Adobe Photoshop software. Adult flies were imaged using a Zeiss Discovery V12 microscope.

Quantitative reverse transcription PCR

For each genotype, three biological and three technical replicates were tested. For each replicate, total RNA was extracted from 50 eye-antennal discs using the RNeasy Plus Mini Kit (Qiagen, 74134) following the manufacturer's protocol. RNA concentration and purity were assessed using a NanoDrop ND-One spectrophotometer (Thermo Fisher Scientific). cDNA synthesis was performed using 100ng of total RNA with the SuperScript III First-Strand Synthesis System (Thermo Fisher Scientific, 11752250) according to the manufacturer's instructions. qRT-PCR was carried out on a LightCycler 96 (Roche) using SYBR Green I Master mix (Roche, 04887352001). Each reaction was performed in a 20 µl mixture containing 10 µl of 2× SYBR Green I Master mix, 2 µl of cDNA and 500 nM of each primer. Primers used to amplify vg and rp49 transcripts were designed using FlyPrimerBank at the Harvard DRSC/TRiP Functional Genomics Resources (vg forward primer 5′-TGCCCCGAAGTTATGTACGG-3′; vg reverse primer 5′-TGGTTGAACCTCTCATACTGGTA-3′; rp49 forward primer 5′-TACAGGCCCAAGATCGTGAA-3′; rp49 reverse primer 5′-TCTCCTTGCGCTTCTTGGA-3′). The thermal cycling conditions included an initial denaturation at 95°C for 5 min, followed by 40 cycles of 95°C for 10 s, 55°C for 10 s and 72°C for 20 s. Melting curve analysis was performed to confirm the specificity of the PCR products. The relative expression levels of the gene of interest (GOI) were calculated using the 2^(−ΔΔCt) method. rp49 was used as an internal control for normalization. Data are presented as the mean±standard deviation (s.d.) of nine independent experiments (three biological and three technical replicates). Statistical significance was assessed using a paired two-tailed Student's t-test, with a P-value <0.05 considered statistically significant.

RNA-seq

Samples were collected from eye-antennal imaginal discs at either 120 h or 144 h AEL. Fifty eye-antennal imaginal discs were dissected from each genotype as previously described (Spratford and Kumar, 2014). Total RNA was extracted using the RNeasy Plus Mini Kit (Qiagen, #74134) per kit protocol, and 0.3ug RNA was used to generate a polyA-strand specific library using the Illumina TruSeq Stranded mRNA library prep kit. Libraries were prepared and sequenced by the Center for Genomics and Bioinformatics (CGB, Indiana University, IN, USA). The read depth for each sample was 15 million reads.

CUT&RUN

Chromatin was extracted and purified from ten third-instar eye-antennal or wing imaginal discs as previously described (Weasner et al., 2023). Samples were treated with rabbit anti-trimethyl-histone H3 (Lys27) (C36B11) (Cell Signaling Technology, #9733, 1:51) and normal rabbit IgG (Cell Signaling Technology, #2729, 1:51) antibodies. Cleaved chromatin fragments were purified using the MinElute PCR Purification Kit (Qiagen, #28004) as per the kit protocol. Libraries were prepared and sequenced by the CGB. The read depth for each sample was 15 million reads.

Computational analysis

RNA-seq

Read quality was assessed using fastqc (v0.11.9) (http://www.bioinformatics.babraham.ac.uk/projects/fastqc/). Genome indices were generated via STAR (Dobin et al., 2013) and were aligned to the dm6 Drosophila genome (with the additional setting --genomeSAindexNbases 13 for Drosophila). Aligned reads were counted using Subread (Liao et al., 2019) function featureCounts with the settings -F GTF -t exon -g gene_id --minOverlap 10 --largestOverlap --primary -s 2 -T. bigWig files of RNA-seq counts were generated using deepTools (Ramirez et al., 2016) (v3.5.1) function bamCoverage with the settings --normalizeUsing CPM -bs 1 --smoothLength 25 --filterRNAstrand [forward or reverse] -p 16. Downstream analysis was performed in Rstudio (v4.2.1). Differential expression analysis was performed using DESeq2 (Love et al., 2014) and heatmaps were generated with ComplexHeatmap (Gu et al., 2016). Differentially expressed transcripts were filtered using the dplyr package (Wickham et al., 2019) to select for enriched/depleted genes in each comparison. Candidate overlap between each time point was then compared with BioVenn web application (Hulsen et al., 2008). GO analysis was performed using clusterProfiler (Yu et al., 2012) eGO function (settings: OrgDb=“org.Dm.eg.db”, ont=“BP”, pAdjustMethod=“fdr”, keyType=“SYMBOL”) or gprofiler2 (Kolberg et al., 2020). Additional scripts for downstream analysis have been deposited in GitHub (https://github.com/brownhe717/Brown_etal_2025).

CUT&RUN

Read quality was assessed with fastqc (v0.11.9) (http://www.bioinformatics.babraham.ac.uk/projects/fastqc/). Genome indices were generated via Bowtie2 (Langmead et al., 2009) and were aligned to the Drosophila dm6 genome and Escherichia coli EB1 genomes (with the settings min_fragment=10 max_fragment=700). The quality and consistency of raw aligned reads were assessed with deeptools (v3.5.1) function multiBamSummary and plotCorrelation (with the settings --whatToPlot heatmap --corMethod spearman --plotNumbers). The number of E. coli aligned reads were accessed using samtools (Danecek et al., 2021) (v1.15.1) command view -c -F 4 and used to generate an appropriate scale factor with the calculation: 10,000/[# E. coli aligned reads]. Normalized bigWig and bedgraph files of Drosophila CUT&RUN reads were generated with deepTools (Ramirez et al., 2016) (v3.9/3.5.1) function bamCoverage with the corresponding --scaleFactor and --outFileFormat [bigwig or bedgraph] setting. Initial visualization of bigWig reads was carried out in Integrative Genomics Viewer (v2.8.0); figures of chromatin tracks were generated using Gviz (Hahne and Ivanek, 2016) in Rstudio (v4.2.1). Additional scripts for downstream analysis have been deposited in GitHub (https://github.com/brownhe717/Brown_etal_2025).

Supplementary Material

Table S1. PcG knockdown phenotypic results.

Table provides the genotype of each gene+c cross and a brief descrip+on of the phenotypic result for each RNAi knockdown experiment conducted in this study.

Table S2. Differential.. expression of knockdown eye-antennal discs.

Table depicts the raw DES eq2 output of each comparison: WT v. toy or PcG RNAi (sheets 1-4), toyKD v. toy-PcG RNAi (sheets 5-7), PcKD v. PcG RNAi (sheets 8-9), PcKD v. toy-PcG RNAi (sheets 10-12), SfmbtKD v. toy-SfmbtKD (sheet 13), and ScmKD v. toy-ScmKD (sheet 14).

Table S3. Common differentially expressed transcripts for each genotype.

Table provides a list of all differen+ally expressed transcripts unique to Pc, Sfmbt, and Scm (sheet 1) or toy-Pc, toy-Sfmbt, and toy-Scm double knockdown discs (sheet 2) as well as those transcripts shared among the three genotypes.