Cut immunostaining also was used to qualitatively confirm the success of our knockdowns. For this we analyzed confocal images of control and knockdown individuals, that were taken with identical imaging settings. For each individual we averaged the gray value from the center of 40 (for the smaller D. melanogaster or 160 (for the larger T. marmoratus) Semper cells and contrasted them to the average of 20 background points for normalization (Supplementary Figures S1C, D). This was possible for T. marmoratus as well as the ct ^GD D. melanogaster line, however knockdown was so efficient in the ct ^V20 line that cells could not be well enough recognized for this analysis.

Control and cut knockdown fly and beetle heads were fixed in 4% formaldehyde overnight at 4°C. These tissues were washed three times with PBS and then cryoprotected overnight at 4°C in sucrose solutions of increasing concentrations (20%, 40%, and 60%). The samples were mounted in Neg50, flash frozen in liquid nitrogen, and cryosectioned at ∼18 µm (Leica CM 1850). The sections were dried, rinsed in PBS, and stained with Phalloidin (following manufacturer’s instruction) and mounted with Fluoromount containing DAPI (Thermo Fisher). Z-stacks for D. melanogaster samples were obtained at a resolution of 1,024 × 1,024 pixels using a Leica Stellaris eight confocal microscope with a ×40 objective. T. marmoratus samples were imaged using a Zeiss LSM 710, AxioObserver confocal microscope with a ×40 objective. The entire eye was tile scanned with a constant pixel size of 0.69 µm. All images were processed using ImageJ, and brightness and contrast were adjusted using Adobe Photoshop 2022.

D. melanogaster and T. marmoratus heads for all groups were dissected in 100% and 50%, respectively, insect ringer (O’Shea and Adams, 1981) to maintain an appropriate osmotic environment and prevent the lens proteins from denaturing. Isolated lenses were suspended in corresponding insect ringer dilutions between coverslips to allow imaging of the undersurfaces of the lenses via DIC microscopy (Nomarski Optics, Olympus BX51 microscope with a ×40 Uplan objective) as well as assessment of the optical quality of the lenses by visualizing the images produced by the lens array. The so-called “hanging drop method” follows protocols that were originally developed by (Homann, 1924) and since then commonly used in investigations of insect optics (Wilson, 1978; Buschbeck et al., 1999; Warrant et al., 2006; Stowasser et al., 2010). A representation of a 2 mm grating was projected through the lenses and pictures were taken of the resulting arrays of images that were focused by the lens arrays. All pictures were obtained using a microscope camera (Qimaging, Retiga 2000R) and the brightness and contrast were adjusted using Adobe Photoshop 2022.

The genetic background in both groups resulted in increased variability in eye color. Since eye color influences the strength of ERG responses, the tested flies were visually color matched. As the two species have different spectral sensitivity peaks (Maksimovic et al., 2011; Sharkey et al., 2020), testing was carried out using a 490 nm light source for D. melanogaster and a 525 nm light source for T. marmoratus. The first test was used to assess the range of light intensities that led to PR responses (V–logI curves). This test consisted of a series of three 1 s long light flashes of increasing intensities, with each flash being ∼10 s apart to allow PR recovery. The second test was used to identify whether the PRs could sustain a response to a train of consecutive light flashes (extended sequence). This test consisted of a series of 150 light flashes at a single light intensity that was within the linear range of PR responses.

For D. melanogaster controls (gfpRNAi and mcherryRNAi) and cutRNAi (ct ^GD and ct ^V20), flies were prepared and used for ERG measurements as previously described (Charlton-Perkins et al., 2017). A V–logI curve was established for each group using the following light intensities (photons/cm²/s): 4.28 × 10¹⁰, 8.15 × 10¹⁰, 2.78 × 10¹¹, 7.20 × 10¹¹, 1.75 × 10¹², 5.42 × 10¹², 1.04 × 10¹³, 5.53 × 10¹³, and 1.07 × 10¹⁴. To ensure steady recording, each series started with a pre-pulse of full light intensity. The extended sequence was administered at 7.20 × 10¹¹ photons/cm²/s (the light intensity at which the PRs were ∼50% saturated), with light and interval durations of 300 ms.

For T. marmoratus controls and cutRNAi, beetles were anesthetized using carbon dioxide. These beetles were secured on a glass slide with dental wax and then allowed to dark adapt for 10 min before establishing V–logI curves using the following light intensities (photons/cm²/s): 5.00 × 10¹¹, 1.90 × 10¹², 5.00 × 10¹², 1.00 × 10¹³, 2.20 × 10¹³, 5.50 × 10¹³, 1.14 × 10¹⁴, 2.25 × 10¹⁴, 3.90 × 10¹⁴, and 1.07 × 10¹⁴. The recording stability was verified through a high intensity pre-pulse. The beetles were then allowed to dark adapt for 10 min before the extended sequence test was performed at 10¹³ photons/cm²/s (the light intensity at which the PRs were ∼50% saturated). As for flies, each light flash had a duration of 300 ms, however stimulus intervals were increased to 500 ms to allow adequate PR recovery.

All the data were analyzed with a custom-made MATLAB code (Riazuddin et al., 2012; Charlton-Perkins et al., 2017). As some groups didn’t show a normal distribution according to the Shapiro-Wilks test, we used Wilcoxon’s rank sum test for all intergroup comparisons (in both species). All graphs were plotted, and statistical tests were calculated in R (version 4.0.3, packages Dplyr, tidyr, and ggplot2).

In D. melanogaster, Cut is expressed in the four SCs of each ommatidium and the interommatidial mechanosensory bristle (Figure 2A; Supplementary Figure S1A) (Blochlinger et al., 1990). Our immunohistochemical investigation using an antibody which was made against the D. melanogaster Cut, found it also expressed in the SCs of T. marmoratus, suggesting a high level of protein conservation between the two species (Figure 2B). N-cadherin and DAPI (for D. melanogaster) or actin and DAPI (for T. marmoratus) were used to confirm the position of the SC cell bodies distal to the developing PRs and rhabdoms. These similarities led us to hypothesize a conserved functional role for this transcription factor between these two eye types. To test for such conserved roles, we used a loss-of-function approach, applying a Semper cell-restricted knockdown in D. melanogaster (Charlton-Perkins et al., 2017) and a generalized dsRNA injection in late stage T. marmoratus larvae (Rathore et al., 2020).

To confirm the SC-restricted knockdown in D. melanogaster, we analyzed whole mounts of 37% developed pupal retina (which is prior to lens formation) in two different cutRNAi lines (ct ^GD and ct ^V20) and compared them to a control line (gfp-RNAi). We found that SC-directed knockdowns reduced Cut expression either partially (ct ^GD; n = 14; Supplementary Figure S1C) or to an undetectable level (ct ^V20; n = 14, Figure 2A) when compared to controls (n = 15). This difference could be due to the driver being heterozygous in ct ^GD flies and homozygous in ct ^V20 flies, or to differences in knock-down efficiencies between these two lines. To verify that cut knockdown was restricted to SCs, we also imaged Cut-expressing interommatidial bristle cells located proximal to the SCs. As expected, Cut expression was unaffected in these cells in all three fly lines (Supplementary Figure S1A).

To ensure a comparable stage and preparation for T. marmoratus, we injected cut dsRNA (Supplementary Figure S2) in late third instar larvae, and used the cross-reacting Cut antibody to detect expression in early developing whole pupal retina. Cut expression was not detectable in any developing ommatidial cell type except for SCs. Consistent with previous findings (Rathore et al., 2020), we found knockdowns in T. marmoratus to be highly successful (Supplementary Figure S1D), with 10 of 12 injected individuals showing reduced Cut expression when compared to controls (n = 8). However, complete cut knockdown was rare, likely because the individuals with the highest knockdown died during early development. This is expected due to cut’s vital role in the development of multiple organ systems in insects (Bauke et al., 2015; Corty et al., 2016; Klann et al., 2021). Accordingly, ∼53% of the cut dsRNA injected individuals died, as opposed to only 27% of the controls. Although the SCs of control individuals strongly expressed Cut, the cut dsRNA-injected individuals exhibited SCs with varying, but greatly reduced Cut expression (Figure 2B).

Although the nature of the knockdown differs in the two species, in both cases, treatments were aimed at detecting SC-driven effects within the ommatidial array. We would like to note that this approach led to differential cut knockdown in the neuropils between the two species, with T. marmoratus cut also being knocked down in other Cut-positive cells in the visual system (e.g., subretinal glia in the lamina), which could lead to additional developmental or neurophysiological phenotypes on photoreceptors in T. marmoratus not detected in D. melanogaster.

To investigate whether cut knockdowns affect the ommatidial organization during development, we examined pupal retinae for patterning defects. In D. melanogaster, N-cadherin allowed visualization of the position of the apical membranes of SCs and their relative position to PRs (Hayashi and Carthew, 2004). For both knockdown lines, we found visible organizational changes in these cell types as well as irregularities in the placement of ommatidia (Figure 2A; Supplementary Figure S1B). Typically, at ∼40% after puparium formation (APF), SCs are expected to be fully differentiated and organized in a characteristic tetrad, with the inner junctions forming an “H”-like pattern (see (Fu and Noll, 1997; Charlton-Perkins et al., 2021) and control in Supplementary Figure S1B). In contrast, the rhabdomeres in PRs are located within the center of each ommatidium and are expected to elongate toward the basement membrane (Longley and Ready, 1995). In both of the ct knockdown lines, the typical “H” shape (indicative of a tetrad of SCs) failed to develop. Instead, an irregular pattern emerged with several instances of triads (Supplementary Figure 1B). In parallel, the cutRNAi of the developing T. marmoratus pupal retina also showed irregularities and instances of SC triads (Figure 2B). Although not verifiable with the N-cadherin antibody due to a lack of cross-reactivity, this phenomenon was clearly apparent with DAPI. The presence of triads in both species suggests that only three of the four SCs in these ommatidia developed properly, with one of the cells likely either absent or radically displaced.

Additionally, for D. melanogaster, control PRs had (as expected) centrally located rhabdomeres, whereas both ct knockdown lines showed displaced rhabdomere organization. Deficiencies included the dislocation of rhabdomeres from the central position of the ommatidium (Figure 2A). In parallel, based on actin counterstaining of the actin-rich rhabdom structures, several of the fused rhabdoms in the T. marmoratus cutRNAi individuals also lost precise alignment at the center of the ommatidia relative to controls (Figure 2B). Taken together, these observations show that cut knockdown in compound eyes results in complimentary developmental organizational deficiencies in both SCs and PRs. To investigate the functional implications of these developmental defects in adult compound eyes, we focused on the three known major functions of SCs: lens formation, PR and rhabdom morphogenesis, and PR functional support.

Lens secretion begins at ∼50% pupal development of D. melanogaster retina, and the role of SCs in this process has been well documented (Cagan and Ready, 1989; Charlton-Perkins and Cook, 2010). Considering our findings that Cut disrupts accurate SC patterning, we next asked if these SC defects could lead to structural and functional defects in adult compound eye corneal lenses. To address this question, we first assessed the outer lens surfaces at an ultrastructural level. In both D. melanogaster (Figures 3A–F) and T. marmoratus (Figures 3G–K), knockdown individuals showed visible defects in lens morphology, shape, and organization when compared to controls.

In D. melanogaster, in contrast to the precisely organized lenses of controls (n = 9; Figures 3A, D), both cut knockdown lines (ct ^GD: n = 10, ct ^V20: n = 14) showed major defects, including a rough eye appearance and variation among lenses (Figures 3B, C). In both lines, we observed instances of fusion of neighboring units and flatter lenses, some of which had visible indentations on the outer lens surface (Figures 3E, F), a phenotype previously referred to as the “blueberry phenotype” (Higashijima et al., 1992). The latter phenomenon was particularly prominent in the ct ^V20 line. The positions of the indentations were often asymmetric, possibly relating to differential contributions by different SCs.

Adult T. marmoratus compound eyes tended to have a smooth surface, which appeared similar in control and cutRNAi individuals at low resolution (Figures 3G, H). However, the RNAi individuals (n = 6) showed an increase in dimple-like developmental defects on the outer eye surface (∼80% of individuals) when compared to controls (∼20%; n = 7; Figure 3I). Higher resolution images revealed that unlike the highly regular and seamlessly connected lens arrays of controls (Figure 3J), cutRNAi individuals generally had irregular lens shapes and pronounced ommatidial boundaries.

To assess lenses from a functional perspective, we isolated lens arrays and mounted them onto a droplet of insect ringer (hanging drop method). This approach allowed examination of the back surfaces of the lenses as well as the visualization of the corresponding images formed by the isolated lens arrays. In both species, control individuals (n = 6 for D. melanogaster and 10 for T. marmoratus) exhibited smooth lens surfaces (Figures 4A–D), and the images produced by the lens arrays were regularly spaced with uniform image magnification (Figures 4A′–D′). In contrast, in D. melanogaster ct knockdown lines, the aforementioned lens deficits were also visible from the back surface (n = 6 for each line; Figures 4B, C). Notably, in ct ^V20 individuals, some lenses appeared necrotic (Iyer et al., 2016), while others showed small indentations. Optically, both lines showed instances where neighboring lenses formed images with irregular placement, focused at different planes, and exhibited variable image magnification (Figure 4B′, C′). For the ct ^V20 line, we also noted that some lenses with holes that entirely failed to form images.

Mirroring the findings from D. melanogaster, cut dsRNA injected T. marmoratus (n = 10) exhibited irregular lens arrays with centrally located indentations on the back surfaces (Figures 4E, F). Many of the lens back surfaces also appeared flatter. Accordingly, the images produced by the lens arrays differed in placement, quality, and image magnification (Figures 4E′, F′). Taken together, these results suggest that Cut is generally required by SCs for accurate lens formation with proper optics in both compound eye types.

SCs are located in close proximity to PRs, and in D. melanogaster, they are known to provide glial support (ionic, metabolic, and structural) to PRs (Charlton-Perkins et al., 2017). The same study also demonstrated that SC-specific pax2RNAi results in defective rhabdom morphogenesis. Considering that Cut functions downstream of Pax2 and based on our observed displacements of the developing rhabdoms, cut knockdown could also lead to misformed adult rhabdoms.

It is important to note that D. melanogaster has open rhabdoms (each consisting of eight rhabdomeres) that extend distally close to the SC bodies (Figures 5A, B), whereas in T. marmoratus, closed rhabdoms are separated from the SCs via a clear zone (Figures 5E, F). As illustrated by phalloidin staining, despite these structural differences, we found similar knockdown phenotypes in the two species.

In D. melanogaster, compared to the pristinely organized rhabdoms of control individuals that traverse the entire length of each ommatidium (n = 10; Figure 5B), the rhabdoms for both ct ^GD (n = 7) Figure 5C and ct ^V20 (n = 6) Figure 5D knockdown individuals were truncated. Generally, they either failed to reach the basement membrane or were present deep in the eye, either traversing the basement membrane or are seen proximally to it. Similarly, in T. marmoratus, controls (n = 7) showed precisely placed PRs with a well-defined clear zone (Figure 5F). However, the retina of cutRNAi individuals (n = 7) showed irregular PR placement (including misplaced nuclei), with some PRs extending to and even through the basement membrane (Figure 5G). Additionally, the clear zone in these individuals was less well defined.

At an ultrastructural level, D. melanogaster controls (based on two TEM preparations) exhibited the classical trapezoid organization (Longley and Ready, 1995) with seven visible open rhabdomeres (Figures 6A, B). In contrast, individuals from both the ct ^GD (n = 3) and ct ^V20 (n = 2) lines exhibited parallel deficiencies, such as irregularly spaced ommatidia (Figures 6C–F), a reduced number of rhabdomeres (Figures 6C–H), split rhabdomeres (Figures 6D, E, G), and fused rhabdomeres (Figure 6H). For T. marmoratus, controls showed regularly organized closed rhabdoms, whereas rhabdom abnormalities were observed in cut RNAi individuals (Figures 6I, J). The retina of cutRNAi individuals showed irregularly placed rhabdoms, with some units being on a different plane within the same array (Figures 6K, N). Additionally, the rhabdoms were frequently incomplete and off center (Figures 6M, O) with instances of split and broken manifestations (Figures 6L, P) and signs of degeneration (Figure 6P). Overall, these results suggest that Cut could function in SCs to direct the proper development of PRs in both compound eye types.

To understand whether morphological defects related to cut knockdowns also affected function, we used electroretinograms (ERGs), an extracellular recording technique that measures the response of the PR array to light stimuli (Belusic, 2011; Riazuddin et al., 2012; Charlton-Perkins et al., 2017).

In D. melanogaster, we first assessed the dynamic ranges of PR responses to increasing light intensities (Figure 7A) and found relatively normal responses in both knockdown lines (Figure 7B). Figure 7A illustrates the response curves for individuals from both control and cutRNAi lines to 490 nm light stimuli of an intermediate light intensity (7.2 × 10¹¹ photons/cm²/s). As expected from a normal fly ERG, control PR responses were flanked by on and off transients. Although flies from both cutRNAi lines had slightly reduced PR responses, there was no statistical difference at two different light intensities (Figure 7C), which may, in part, be due to the high variability in all groups, which may be related to slight differences in eye color. In T. marmoratus, all controls had normal PR responses, but dramatic differences existed in cutRNAi individuals, which were separated into three categories (Figure 7D): 1) Individuals with normally shaped responses at all light intensities, 2) individuals with normally shaped responses at low light intensities but inverted responses at higher light intensities, and 3) individuals with reversed-polarity responses at all light intensities. To keep the quantitative analysis comparable to controls, only individuals with electronegative responses were incorporated into the statistical analysis. Figure 7E illustrates such normal and reversed potential responses. In some cases, in a train of pulses, only the first response was inverted (Figure 7H). As shown in Figure 7F, even the normal, electronegative responses of cutRNAi individuals tended to be lower than those of controls (n = 15), with a significant difference at a light intensity of 10¹³ photons (n = 7; p = 0.0015, Wilcoxon’s test) but not at 4 × 10¹³ (n = 4; p = 0.08, Wilcoxon’s test) (Figure 7G). Note that statistical power was lost at higher light intensities due to smaller sample sizes resulting from an increased occurrences of reversed polarity potentials (5.00 × 10¹¹, n = 8; 1.90 × 10¹², n = 8; 5.00 × 10¹², n = 8; 1.00 × 10¹³, n = 7; 2.20 × 10¹³, n = 5; 5.50 × 10¹³, n = 5; 1.14 × 10¹⁴, n = 5; 2.25 × 10¹⁴, n = 4; 3.90 × 10¹⁴, n = 5; and 1.07 × 10¹⁴, n = 4).

It has previously been demonstrated that certain genetic perturbations can affect the ability of flies to maintain a proper photoresponse throughout a long series of light pulses (Riazuddin et al., 2012). To test if such deficiencies exist in cut knockdowns, we exposed flies and beetles to a series of 150 light pulses (extended sequence) (Supplementary Figure S3). To assess differences in the PR potentials at earlier and later time points, the responses were analyzed by grouping the data into 10-point bins. Since some early signal reduction is normal (due to adaptation), we limited our analysis to bins 3–15. No significant differences between these bins were observed in any of the fly lines (Supplementary Figures S3A, B, p = 0.97, 0.97, 0.44, and 0.58, Wilcoxon’s test) or the controls and cut RNAi beetles (Supplementary Figures S3C, D, p = 0.29 and 0.28, Wilcoxon’s test). Consistent with our analysis at different light intensities (Figure 7G), cut RNAi beetles showed slightly lower responses than the controls. Apart from some inverted responses in cut RNAi injected beetles (Figures 7E–H), our ERG analysis didn’t reveal any major differences between the controls and knockdowns in either compound eye type.

Our comparative study suggests that if cut is knocked down, the crystalline precision of the eye is disrupted relatively early in development, irrespective of the compound eye type. The key to this phenomenon likely lies in the central role of SCs for ommatidium development and of Cut for proper patterning of SCs.

During eye development, SCs undergo dramatic changes in organization (Cagan and Ready, 1989) and play important roles in recruiting later-developing PPCs (Nagaraj and Banerjee, 2007) and regulating the orientation of developing PRs. The specific patterning steps of SCs likely include a combination of forces driven by the actomyosin cytoskeleton, cell adhesion (Cadherins and Nephrins), endocytosis, and Notch signaling (Chan et al., 2017; Blackie et al., 2020; Blackie et al., 2021; Charlton-Perkins et al., 2021). As Cut expression is detectable shortly after SC specification (Blochlinger et al., 1990; Canon and Banerjee, 2003; Charlton-Perkins et al., 2011), it is well positioned to contribute to the regulation of these patterning events. Consistent with this possibility, in both species, cut knockdowns lead to defects in cell placement, including the presence of triads or displaced tetrads in the SC layer (Supplementary Figure S1B) and the lateral displacement of adjacent PRs (Figure 2A). Interestingly, these defects are reminiscent of those observed in the developing D. melanogaster mutant pupal retina for Pax2, Hibris, Roughest, Mastermind (a Notch signaling inhibitor), and Wingless (Grillo-Hill and Wolff, 2009; Cordero and Cagan, 2010; Charlton-Perkins et al., 2011; Blackie et al., 2021). Although it is well established that Cut interacts with wingless and Notch signaling in the wing disc (Micchelli et al., 1997), evidence suggests that it doesn’t interact with wingless in the eye disc (Cordero and Cagan, 2010). Cut’s interactions with other genes in the retina therefore remain subject of further studies.

In cut knockdowns, we observed severe abnormalities in lens formation, with many parallel deformities in the two species that suggest a conserved overall function. Some differences, such as defects on the outer lens surfaces, are likely related to eye-type-specific differences in lens organization. Specifically, D. melanogaster eyes are characterized by biconvex lenses, whereas T. marmoratus eyes are characterized by plano convex lenses (Figure 1). These differences may have caused the fly eyes to show rough lens surfaces with prevalent holes (blueberry phenotype), whereas the beetle eyes maintained relatively smooth outer surfaces with only occasional dimples (Figure 3). However, knockdowns of cut in both species exhibited indentations on the inner lens surfaces as well as a generally disorganized lens arrays containing misshaped and fused lenses. As the inner surface is closer to the location of lens secretion (a process mediated by SCs and all pigment cells (Cagan and Ready, 1989; Wang et al., 2012; Chaturvedi et al., 2014; Stahl et al., 2017b), it isn’t surprising that inner-surface deficits are more consistent between the two eye types. Similar accessory-cell based lens defects have already been observed in several D. melanogaster eye mutants (Higashijima et al., 1992; Charlton-Perkins et al., 2011; Wang et al., 2022). Of specific interest are the rough eye mutants of prospero and pax2, which specify SCs combinatorially (Charlton-Perkins et al., 2011), and bar, which is required for PPC function (Higashijima et al., 1992). Both these mutants and our knockdowns have irregularly shaped lens arrays, individual lenses with holes (including the blueberry phenotype), and flattened and fused lenses. These commonalities highlight the complexity of the process of proper lens formation, which involves many genes that act synergistically in multiple cell types.

Based on the severity of lens defects observed in both cut RNAi species, it is not surprising that we also noted major optical defects. In both species, lenses failed to form sharp images or to focus on the same plane, with associated variability in image magnification (Figure 4). Such optical assessments are a powerful but underutilized method for characterizing the functional deficits of lenses, although morphometric modeling (Wang et al., 2022) has been implemented as an alternative assessment method. The observed optical deficits likely have dramatic implications on the spatial resolution of both compound eye types. It remains an open question whether cut knockdowns also affect the optics of the crystalline cone (CC), another important extracellular optical component of superposition eyes (Warrant and McIntyre, 1993; Meece et al., 2021). The CC, which re-inverts the image to allow for an upright image at the level of PRs, is likely also formed (or at least contributed to) by SCs (Nilsson, 1989). Unfortunately, unlike in other beetles, such as fireflies, in T. marmoratus, the CCs aren’t retained when isolating lenses and hence cannot be optically assessed.

Although Cut affects SCs, cutRNAi individuals of both species also exhibit severe morphological defects in adjacent PR cells. On a gross morphological level, this includes laterally displaced rhabdoms that are already apparent in early development (Figure 2) as well as longitudinally displaced PR nuclei, shortened rhabdoms, and rhabdoms that extend through the basement membrane (Figure 5). In D. melanogaster, the latter two phenotypes are particularly severe. Notably in that species the orientation of rhabdomeres turns during development (Ready, 2002). It is possible that observed phenotypes relate to perturbations of that process, a possibility that warrants further investigation. These observed differences could be related to variations in the knockdown severity but could also be related to organizational differences between the two eye types, with rhabdoms in beetles originating much deeper in the eye but terminating more distal to the basement membrane than those in flies. Parallel defects were also observed in cross-sections at the ultrastructural level, with rhabdoms of both species showing a variety of morphological defects including splitting, variable sizing, and lateral displacements (Figure 6). In D. melanogaster, where the open rhabdom facilitates checking for the presence of all rhabdomeres, a few ommatidia appear to be missing some rhabdomeres. However, they could also have been displaced to a different plane; evaluation of this possibility requires future 3D TEM analysis. Our data is consistent with a growing body of evidence that raises the possibility that an intact SC–PR interface is necessary for proper PR placement (during early pupation) and rhabdom elongation (during mid to late pupation (Longley and Ready, 1995). It is noteworthy that SC-specific pax2 RNAi flies have very similar patterning defects (Fu and Noll, 1997; Charlton-Perkins et al., 2017), as might be expected if these two genes act through a common pathway. In contrast, Blimp-1 knockdown in SCs also leads to eyes with shortened rhabdoms (Wang et al., 2022), although they do not cross the basement membrane. Interestingly, even defects in surrounding support cells, such as PPCs, have been reported to result in similar rhabdom deficiencies in flies, seen in Bar mutants (Higashijima et al., 1992). Overall, these findings indicate that the accurate patterning of SCs is necessary to allow PRs to be properly shaped and positioned. It has already been demonstrated that ineffective cell adhesion in PRs is a major contributor (Longley and Ready, 1995; Izaddoost et al., 2002), a topic that warrants further exploration.

In line with our interpretation that structural PR deficits in cut knockdown flies and beetles are secondary to disturbances of the entire ommatidial unit, we found that the physiological response of PRs was relatively intact, although there were notable phenotype differences between the two species. In D. melanogaster, the cutRNAi flies had normal responses despite abnormal morphology, similar to SC-specific pax2 knockdown flies (Charlton-Perkins et al., 2017). The possible physiological phenotypes in flies are worth further investigation, as our study was hampered by a relatively large variation in responses, likely due to minor differences in eye color, and more subtle physiological deficits may only become apparent when PRs are substantially challenged (Riazuddin et al., 2012; Charlton-Perkins et al., 2017).

In contrast to flies, the PR responses in cutRNAi beetles were significantly lower than those in the controls for some light intensities and were even inverted in many cases (Figure 7). This seemingly reversed phenotype is reminiscent of that of repo (a general glial marker) mutant flies (Xiong et al., 1994). Since in contrast to D. melanogaster, the effect of RNAi in T. marmoratus wasn’t cell type specific, it is conceivable that the observed species-specific differences are related to cut-dependent roles in other cell types such as the subretinal glia (Bauke et al., 2015). Interestingly, the reversed response was sometimes transient, being present only in the first pulse (Figure 7). As the ERG is a field potential measured from the surface of the eye (Belusic, 2011), this waveform could be related to defects in the resistance barrier and the complex flow of currents in the relatively tight spaces of the eye rather than an altered PR response.

The data presented here identify cut as part of a support-cell-specific conserved gene network that is essential for proper eye development in different compound eye types, with largely parallel phenotypes in cut knockdowns in flies and beetles. Interestingly, several defects observed here parallel those in SC-pax2 mutants, thus providing supportive functional evidence for a pax2-cut model in SCs that is essential for proper cell adhesion, lens secretion, and PR morphology (Figure 8). Here, we studied these knockdown phenotypes for the first time within the framework of a direct comparison between eye types, laying the groundwork for further investigations into this model and how it might regulate the development of different compound eyes. Since diverse compound eyes, including those with different optics, are composed of ommatidia with highly conserved cellular compositions (Paulus, 1979; Nilsson, 1989), it is not surprising that the disruption of a central cell type, such as SCs, has dramatic effects on the entire complex ommatidial unit.

However, the manner in which related deficits manifest could be, at least to some degree, eye type specific. For example, we expect deficits to be particularly detrimental for the superposition optics of beetles, as a high level of precision in organization is required for proper function. Our data suggest that cutRNAi in beetles destabilizes this precision in multiple ways, including affecting ommatidial array regularity, lens integrity and optics, clear zone integrity, and PR placement. As each of these factors could itself be detrimental to proper function, we expect that even a relatively mild cut knockdown phenotype would lead to relatively dramatic visual deficits, with the potential of completely losing the ability to resolve images. For D. melanogaster eyes, the loss of spatial resolution is also expected, but this effect may be more subtle, as in parts of the eye, individual lenses would continue to project images (albeit possibly blurry images) on the corresponding PRs. Additionally, both eye types are typical for flying insects that capitalize on precise sampling from neighboring units for proper motion computation, which generally relies on elementary motion detectors (for a review, see (Egelhaaf et al., 1988). Hence, even subtle phenotypes are expected to lead to specific deficits, which is an interesting area for further investigation.

Although this study compared two optically relatively different compound eye types, we have only scratched the surface of how Cut may be involved in the incredible diversity of eyes that have evolved in the lineage of compound eyes (Buschbeck, 2014; Morehouse et al., 2017; Lavin et al., 2022) and are composed of a set of highly conserved cell types (Paulus, 1979). One particularly intriguing observation in our study is the occurrence of fused lenses, which are particularly significant from an optical perspective, as such fusion events may give rise to eye formations in which one lens serves multiple PRs, the hallmark of a single-chamber, image-forming eye. Transitions from compound eyes to image-forming eyes have been observed in nature (for example, in mysid shrimp (Nilsson and Modlin, 1994) but have thus far remained unexplored from a genetic perspective. We anticipate that our study is part of the groundwork for further explorations on support-cell-mediated eye diversification and how deeply conserved eye development genes beyond the RDGN contribute to the manifestation of optically highly diverse eye types.