A protein-trap allele reveals roles for Drosophila ATF4 in photoreceptor degeneration, oogenesis and wing development

Virtually all organisms have evolved stress response mechanisms to mitigate the impact of homeostatic imbalance. The integrated stress response (ISR) pathway, conserved from yeast to humans, is initiated by a collection of stress-responsive kinases. The ISR pathway has been linked to the etiology of a number of human diseases, including neurodegenerative disorders, diabetes and atherosclerosis, among others (Back et al., 2012; Chan et al., 2016; Ivanova and Orekhov, 2016; Ma et al., 2013). Thus, there is immense interest in identifying specific ISR signaling factors and their roles in these pathologies.

Each ISR kinase responds to a different type of stress: PERK (also known as EIF2AK3), an ER-resident kinase, responds to disruption in endoplasmic reticulum (ER) homeostasis (e.g. misfolding proteins and calcium flux); GCN2 (also known as EIF2AK4), a cytoplasmic kinase, responds to amino acid deprivation; PKR (also known as EIF2AK2), a cytoplasmic kinase, responds to double-stranded RNA; and HRI (also known as EIF2AK1), a cytoplasmic kinase, responds to oxidative stress (Donnelly et al., 2013). More recently, MARK2 has been identified as an additional eIF2α kinase that responds to proteotoxic stress (Lu et al., 2021). When activated by the corresponding cellular stress, the ISR kinases phosphorylate the same downstream target: the α-subunit of the initiator methionyl-tRNA (Met-tRNA_i^Met) carrying complex, eIF2. Such phosphorylation of eIF2α leads to decreased availability of Met-tRNA_i^Met, resulting in lowered cellular translation (Sonenberg and Hinnebusch, 2009). However, the translation of some mRNAs with unusual 5′ leader arrangements, such as the one encoding the ISR transcription factor ATF4, is induced even under such inhibitory conditions (Hinnebusch et al., 2016). ATF4 is a bZIP (basic leucine zipper) transcription factor that induces the expression of stress response genes, including those involved in protein folding chaperones, amino acid transporters and antioxidant genes (Back et al., 2009; Fusakio et al., 2016; Han et al., 2013; Shan et al., 2016).

The number of ISR kinases varies depending on organismal complexity, e.g. GCN2 in Saccharomyces cerevisiae (yeast), GCN2 and PERK in Caenorhabditis elegans (worms) and Drosophila melanogaster (flies), and additional ISR kinases in Danio rerio (zebrafish) and other vertebrates (Mitra and Ryoo, 2019; Ryoo, 2015). Although these kinases induce a few downstream transcription factors (Andreev et al., 2015; Brown et al., 2021; Palam et al., 2011; You et al., 2021), ATF4 remains the best characterized (Donnelly et al., 2013). Drosophila has a functionally conserved ortholog referred to as cryptocephal (crc) (Fristrom, 1965; Hewes et al., 2000). In addition to its well-characterized roles during cellular stress, a plethora of studies have demonstrated roles for ISR signaling components during organismal development (Mitra and Ryoo, 2019; Pakos-Zebrucka et al., 2016). In Drosophila, loss of Gcn2 results in decreased lifespan and increased susceptibility to amino acid deprivation and bacterial infection (Kang et al., 2017; Vasudevan et al., 2017). Drosophila Perk is highly expressed in the endodermal cells of the gut during embryogenesis, and has also been demonstrated to regulate intestinal stem cells in adults (Wang et al., 2015). Although both Gcn2 and Perk Drosophila mutants survive to adulthood (Kang et al., 2017; Vasudevan et al., 2020), mutations in crc result in significant lethality during larval stages. The crc hypomorphic point mutant crc¹, which causes a single amino acid change, results in delayed larval development and subsequent pupal lethality (Fristrom, 1965; Hewes et al., 2000; Vasudevan et al., 2020). The most striking phenotype of the crc¹ mutants is the failure to evert the adult head during pupariation, along with failure to elongate their wings and legs (Fristrom, 1965; Gauthier et al., 2012; Hewes et al., 2000; Vasudevan et al., 2020).

The larval and pupal lethality of known crc alleles have limited our understanding of its role in adult tissues. Additionally, study of the role of Crc using mitotic clones has been impeded by the cytogenetic proximity of crc to the widely used flippase recognition target (FRT)40 element. Here, we report that a GFP protein-trap reporter allele in the crc locus acts as a hypomorphic mutant that survives to adulthood. We use this allele to discover that loss of crc results in higher rates of retinal degeneration in a Drosophila model of autosomal dominant retinitis pigmentosa (adRP), a human disease with an etiology linked to ER stress. Adult crc mutants show increased susceptibility to amino acid deprivation, consistent with what was previously known for GCN2. Additionally, we observe several developmental defects in adult tissues, including reduced female fertility due to a block in oogenesis. We also observe wing vein defects and overall reduced wing size in both male and female crc mutants.

In seeking endogenous reporters of Crc activity, we examined a ‘protein trap’ line for crc, generated as part of the Gene Disruption Project (Nagarkar-Jaiswal et al., 2015a,b; Venken et al., 2011). The Gene Disruption Project is based on a Minos-mediated integration cassette (MiMIC) element inserted randomly into various regions in the Drosophila genome. The cassette can be subsequently replaced with an EGFP-FlAsH-StrepII-TEV-3xFlag (GFSTF) multi-tag cassette using recombination-mediated cassette exchange. One such insertion recovered through this project is in the intronic region of the Drosophila crc locus, which has been subsequently replaced with an GFSTF multi-tag cassette (Fig. 1). The splice donor and acceptor sequences flanking the cassette ensure that the GFSTF multi-tag is incorporated in the coding sequence of most crc splice isoforms to generate a Crc fusion protein (Fig. 1). This crc reporter allele is henceforth referred to as crc^GFSTF, with the encoded fusion protein referred to as Crc-GFP.

Our laboratory and others have utilized acute misexpression of Rh1^G69D, an ER stress-imposing mutant protein, in third instar larval eye disc tissues, using a GMR-Gal4 driver (GMR>Rh1^G69D), as a facile method to activate the Perk-crc pathway (Kang et al., 2015, 2017; Ryoo et al., 2007). We tested the utility of the crc^GFSTF allele as an endogenous reporter for Crc levels, and found robust induction of Crc-GFP in third instar larval eye discs in response to misexpression of Rh1^G69D protein, but not in response to control lacZ protein in the crc^GFSTF/⁺ background (Fig. 2A,B). To validate that such induction was downstream of PERK activation, caused by the misexpression of Rh1^G69D, we generated Perk mutant FRT clones negatively marked by DsRed expression in the glass multiple reporter (GMR) compartment using ey-FLP. Although control clones showed no change in induction of Crc-GFP (Fig. 2C), Perk^e01744 mutant clones showed a complete loss of Crc-GFP in GMR>Rh1^G69D eye imaginal discs (Fig. 2D). We also validated these observations in whole-animal Perk^e01744 mutants, in which we observed a complete loss of Crc-GFP in GMR>Rh1^G69D eye imaginal discs (Fig. S1).

To examine whether Crc-GFP could also be used as a readout for GCN2 activation, we dissected fat bodies from wandering third instar larva, in which we have previously reported GCN2-dependent Crc activation (Kang et al., 2015, 2017). We observed Crc-GFP signal localized to the nucleus in the larval fat bodies (Fig. S2). Such signal was substantially depleted using fat body-specific RNAi knockdown of ATF4 or GCN2 (Fig. S2).

As the induction of Crc in response to stress is regulated at the level of mRNA translation, we wanted to ensure that the induction of Crc-GFP we observed in Fig. 2A-D is reflective of translation regulation via the crc 5′ leader sequence. The crc 5′ leader is structured such that the main open reading frame of crc is favorably translated when eIF2 availability is reduced, such as during phospho-inactivation of eIF2α by ISR kinases (Hinnebusch, 1984; Kang et al., 2015). It has also been previously demonstrated that phosphorylation of a subset of cellular eIF2α is sufficient to diminish initiator methionine availability, thus mimicking ISR activation (Ramaiah et al., 1994). To imitate reduction of eIF2 availability by ISR kinases, we generated a phospho-mimetic transgenic line in which the Ser51 in eIF2α is mutated to Asp51 (UAS-eIF2α^S51D). We also generated a corresponding control transgenic line containing wild-type eIF2α (UAS-eIF2α^WT). We used this genetic mimetic of ISR activation to test whether crc^GFSTF reported crc translation induction under reduced eIF2 availability conditions. While GMR>eIF2α^WT discs showed no detectable levels of Crc-GFP, we found that GMR>eIF2α^S51D led to robust induction of Crc-GFP in eye discs, as detected by immunostaining with anti-GFP (Fig. 2E,F). These data demonstrate the applicability of crc^GFSTF as a reliable reporter of endogenous Crc expression downstream of ISR activation.

Similar to the previously characterized crc hypomorphic mutant allele crc¹ (Fristrom, 1965; Hewes et al., 2000), we observed that flies homozygous for crc^GFSTF exhibited a delay in head eversion and showed anterior spiracle defects. However, unlike the crc¹ mutants, none of the crc^GFSTF homozygotic pupae exhibited complete loss of head eversion, indicating that crc^GFSTF is likely a weaker hypomorphic allele than crc¹. To further assess the effects of the crc^GFSTF allele, we performed lethal phase analysis of development, starting at the first instar larva. We found that a little over 50% of crc^GFSTF homozygotes were larval lethal (Fig. 3A), which is remarkably similar to the larval lethality we previously reported for crc¹ (Vasudevan et al., 2020). However, unlike crc¹ homozygotes, only a small percentage of crc^GFSTF homozygotes showed prepupal and pupal lethality, with ∼30% of animals eclosing as adults (Fig. 3A). To ensure that these developmental defects cannot be attributed to background mutations in crc^GFSTF, we performed lethal phase analysis on crc^GFSTF in transheterozygotic combinations with the hypomorphic crc¹ allele. We found that crc^GFSTF/crc¹ transheterozygotes showed similar levels of larval and pupal lethality to crc^GFSTF homozygotes, with ∼25% of animals surviving to adulthood (Fig. 3A). These data together suggested that the crc^GFSTF allele may function as a crc loss-of-function allele.

To examine whether crc transcript levels are affected in crc^GFSTF mutants, we performed qPCR in the wandering third instar larval stage, when Crc activity is known to be high in fat tissues (Kang et al., 2015, 2017). We found that crc^GFSTF homozygotes showed a ∼65% decrease in crc transcript levels in comparison to control animals (Fig. 3B). We also tested Crc activity by measuring mRNA levels of the well-characterized Crc transcriptional target 4E-BP (Drosophila Thor). We observed ∼40% lower levels of Thor in crc^GFSTF in comparison to control animals (Fig. 3B). This reduction in transcript levels of Crc targets was also reproducible in crc^GFSTF/crc¹ transheterozygotes (Fig. 3B). Taken together, these data indicate that crc^GFSTF acts as a mild hypomorphic mutant allele of crc.

Nearly 30% of all adRP mutations are found in the rhodopsin gene (Illing et al., 2002; Kaushal and Khorana, 1994). Several of these Rhodopsin mutations result in misfolding proteins, which inflict ER stress (Kroeger et al., 2019). However, the role of ATF4 in adRP has remained unclear. We sought to resolve this using the crc^GFSTF allele in a Drosophila model of adRP.

Clinically, adRP is characterized by age-related loss of peripheral vision, resulting in ‘tunnel vision’ and night blindness, due to degeneration of rod photoreceptors (Kaushal and Khorana, 1994). The Drosophila genome encodes several Rhodopsin genes, including ninaE, which encodes the Rhodopsin-1 (Rh1) protein. The ninaE^G69D mutation captures essential features of adRP etiology: flies bearing one copy of the dominant ninaE^G69D allele exhibit age-related retinal degeneration, as seen by photoreceptor cell death (Colley et al., 1995; Kurada and O'Tousa, 1995). This mutant encodes a protein with a negatively charged Asp residue in the transmembrane domain; therefore, it is predicted to disrupt the folding properties of Rh1. Consistently, the ninaE^G69D mutant activates ER stress markers in photoreceptors (Ryoo et al., 2007). More recent gene expression profiling experiments found that ninaE^G69D/⁺ photoreceptors induce many ISR-associated genes, including crc itself (Huang and Ryoo, 2021).

We found that crc^GFSTF/crc¹; ninaE^G69D/⁺ animals exhibited rapid retinal degeneration in comparison to crc^GFSTF/⁺; ninaE^G69D/⁺ control animals, as monitored by pseudopupil structures in live flies over a time course of 30 days (Fig. 4A). The earliest timepoint when control animals exhibit retinal degeneration is typically around 13-15 days; however, crc homozygous mutant animals exhibited onset of retinal degeneration as early as 4 days, with all animals displaying complete loss of pseudopupil structures by day 14 (Fig. 4A). Further analysis of photoreceptor integrity by actin immunostaining following dissection showed that even young (2 days old) crc^GFSTF/crc¹; ninaE^G69D/⁺ flies showed evident disruption of ommatidial organization in comparison to ninaE^G69D/⁺ animals, in which ommatidial organization was relatively unperturbed (Fig. 4B-D). At day 7, the majority of ninaE^G69D/⁺ animals showed intact pseudopupils and identifiably regular ommatidial arrangements (Fig. 4A,E). In contrast, we observed considerable disruption of ommatidial structures in both crc¹ and crc^GFSTF/crc¹ mutants bearing the ninaE^G69D allele. (Fig. 4A,F,G). Interestingly, we also found that crc^GFSTF/crc¹ animals exhibited age-dependent retinal degeneration even in the absence of ninaE^G69D, indicating that age-related physiological decline requires a protective role for Crc in photoreceptors (Fig. 4A).

To measure the expression of Crc in aging photoreceptors, we performed western blotting of adult fly heads from young (2 days old) and old (14 days old) flies to detect Crc-GFP. Although young control flies (crc^GFSTF/⁺) showed very low levels of Crc-GFP, flies bearing one copy of ninaE^G69D showed a substantial induction of Crc-GFP (Fig. 4H,I). We observed that Crc-GFP increases with age in 14-day-old control flies (crc^GFSTF/⁺), with a concomitant increase in Crc-GFP in ninaE^G69D/⁺ flies (Fig. 4H,I). These data substantiate the role of Crc in photoreceptors suffering from ER stress induced by misfolding-prone Rh1^G69D, thus providing a basis for the protective roles of Perk in retinal degeneration.

In addition to rendering a protective effect during ER stress inflicted by Rh1^G69D, we also tested whether Crc had an effect during amino acid deprivation in adult animals. We tested this by subjecting crc^GFSTF/crc¹ animals to amino acid deprivation by rearing animals on 5% sucrose-agar. Although a majority of control animals survived up to 8 days, crc^GFSTF/crc¹ animals steadily succumbed to amino acid deprivation, starting at day 2 with no survivors by day 6 (Fig. 4J). This is consistent with the idea that Crc mediates the GCN2 response to amino acid deprivation in adult Drosophila.

crc^GFSTF provided an opportunity to examine previously unreported roles for crc in adult flies. We first observed that wings from both crc^GFSTF homozygotes and crc^GFSTF/crc¹ transheterozygotes showed a range of venation defects (Fig. 5A-C). The Drosophila wing has five longitudinal veins (annotated L1-L5) and two cross veins, anterior and posterior, labeled ACV and PCV, respectively (Fig. 5A). Severe wing defects in crc^GFSTF homozygous flies were characterized by ectopic venation on L2, between L3 and L4, on L5, and ectopic cross veins adjacent to the PCV (Fig. 5B,B′). crc^GFSTF/crc¹ transheterozygotes largely showed milder wing defects, characterized by ectopic venation on the PCV and on L5 (Fig. 5C,C′). We quantified these wing phenotypes in over 40 animals of each sex, and found that the penetrance and severity of the phenotype was much stronger in females than in males (Fig. 5E). To ensure that the phenotypes were not due to background mutations, we performed a genomic rescue experiment using a BAC-clone based chromosomal duplication covering the crc locus, Dp (90599). We found venation phenotypes in crc^GFSTF homozygotic mutants to be substantially, albeit incompletely, rescued by Dp (90599) (Fig. 5D-E).

We also observed that crc mutant wings were smaller than in control animals (Fig. 5A-C). Quantification of the wing area revealed a statistically significant decrease in wing blade size in crc^GFSTF and crc^GFSTF/crc¹ (Fig. 5F). To exclude the possibility of dominant negative effects of crc^GFSTF, we also tested wings from crc^GFSTF/⁺ heterozygotes but found no wing defects in these animals (Fig. S3). We were unable to detect Crc-GFP expression in the developing wing discs. This may be because of insufficient sensitivity of the reporter, or possibly indicates a non-autonomous role for crc in wing development. It is notable that Gcn2 depletion in the wing reportedly causes venation (Malzer et al., 2018). Thus, our results suggest that Gcn2-mediated Crc activation contributes to proper wing vein development.

In trying to establish a stock of crc^GFSTF, we observed that when mated to each other, crc^GFSTF homozygotic males and females produced no viable progeny, with very few of the eggs laid hatching to first instar larvae. To determine whether this loss of fertility in crc^GFSTF is due to loss of fertility in males, females or both, we separately mated crc mutant females to healthy control (genotype; yw) males and vice versa. We observed that although crc^GFSTF and crc^GFSTF/crc¹ males produced viable progeny at similar rates to control yw males (data not shown), crc mutant females showed ∼50% reduction in egg laying compared to control females (Fig. 6A), again with very few of the eggs laid hatching to first instar larvae. Upon closer observation, we saw defects in the dorsal appendages of eggs laid by crc mutant females, from mild phenotypes, such as the shortening of the appendages, to a complete absence of one or both appendages (Fig. 6B). The proportion of eggs showing such dorsal appendage defects were significantly higher in crc mutants than in control yw animals. Both the overall fertility defect and dorsal appendage defects in crc mutants were significantly rescued with the introduction of Dp(90599) (Fig. 6A).

Dorsal appendages are specified and develop in the final stage of oogenesis. Each Drosophila ovary comprises 14-16 developing follicles called ovarioles, with germline stem cells, residing at the anterior apex, undergoing differentiation along the ovariole in individual egg chambers (Lobell et al., 2017). Each egg chamber represents a distinct stage in ovulation, with stage 14 representing a mature egg (see schematic in Fig. 6C). To examine whether the dorsal appendage defects resulted in decreased fertilization of eggs, we measured fertilization rates of laid eggs by mating control and crc mutant females with don juan-GFP (dj-GFP) males (Santel et al., 1997). Dj-GFP marks individual spermatids, which can be observed in fertilized eggs under a fluorescent microscope. Our analysis showed no significant change in rates of fertilization between eggs laid by yw, crc^GFSTF and crc¹/crc^GFSTF females (Fig. 6D). These data indicate that the fertility defects in crc mutants are due to loss of Crc function in female flies.

To further dissect the fertility defects, we examined ovaries from crc mutant animals. We observed that ovaries from crc^GFSTF and crc^GFSTF/crc¹ were considerably swollen compared to control ovaries (Fig. S4A). Several ovarioles within crc mutant ovaries showed enlarged stage 10 egg chambers, indicative of a stall in oogenesis (white arrowheads in Fig. S4A). Indeed, examination of individual ovarioles from crc mutant ovaries counterstained for actin showed that loss of crc results in an abnormal arrangement of early stage egg chambers (Fig. 6E,F). Although ovarioles from control animals showed sequentially staged and spaced egg chambers culminating in mature stage 14 eggs (Fig. 6C,E), ovarioles from crc^GFSTF and crc^GFSTF/crc¹ appeared to be arrested at stage 10, with improper spacing between egg chambers in earlier stages (white arrowheads, Fig. 6F,G). We quantified the number of ovarioles that displayed such arrest and found that more than half of crc mutant ovarioles (∼9) in each ovary showed stage 10 arrest compared to an average of 2-3 ovarioles arrested in ovaries from corresponding control animals (Fig. 6H).

To determine whether the arrested egg chambers underwent subsequent cell death, we immunostained ovaries with an antibody that detects proteolytically activated (cleaved) caspase Dcp-1 (Vasudevan and Ryoo, 2016). We observed that stage 7/8 egg chambers from several crc^GFSTF and crc^GFSTF/crc¹ ovarioles showed strong cleaved Dcp-1 staining (Fig. 7A-C). Analysis from over ten young animals (2 days old) indicated that at least one ovariole in each crc mutant ovary showed strong cleaved Dcp-1 staining in stage 7/8 egg chambers, in stark contrast to none in control ovaries (Fig. 7D). These data suggest that the decrease in fertility in crc^GFSTF and crc^GFSTF/crc¹ females is correlated with cell death in stage 7 and 8 egg chambers during oogenesis.

Reduced fertility has also been demonstrated to be a consequence of dysregulation of stalk cells, which connect the egg chambers of an ovariole (Fig. 6C; Borensztejn et al., 2018). Specifically, either a failure to reduce stalk cell numbers by apoptosis during development, or excessive culling of stalk cells, were shown to decrease fertility (Borensztejn et al., 2018). To examine whether the fertility defects in crc mutants could be attributed to dysregulation in stalk cell numbers, we stained control and mutant ovaries with a stalk cell marker, castor (Chang et al., 2013). We observed that although ovarioles from control egg chambers showed ∼7 stalk cells between their stage 5/6 and 7/8 egg chambers, crc mutants showed significantly fewer stalk cells (Fig. 7E-H). These data suggest that dysregulation of stalk cell apoptosis may contribute to the fertility defects seen in crc mutants.

To examine which cell types express Crc in the ovary, we immunostained ovaries with GFP antibody to detect Crc-GFP. However, we were unable to detect Crc-GFP with anti-GFP (Fig. S4B,C), suggesting that Crc may regulate ovulation non-autonomously. We also attempted western blotting of ovary extracts to detect Crc-GFP but did not observe any detectable signal (data not shown). A previous study had suggested a non-autonomous role for fat body Gcn2 in the regulation of oogenesis (Armstrong et al., 2014). Consistent with these observations, we detected high levels of Crc-GFP fusion protein in adult abdominal fat tissues from crc^GFSTF animals (Fig. S5A,B). Further, using a fat body-specific driver to deplete crc in adult fat tissues led to decreased egg laying (Fig. S5C), similar to the phenotypes observed in crc mutants (Fig. 6A). These data raise the possibility that Crc mediates Gcn2-signaling in fat tissues to non-autonomously regulate oogenesis.

ISR signaling is associated with various pathological conditions, but the role of Drosophila crc in adult tissues had remained unclear. This may be partly because the cytogenetic location of crc is very close to FRT40, and therefore, attempts to study Crc function using conventional genetic mosaics have been unsuccessful. Our understanding of the role of Crc in adult Drosophila tissues thus far has entirely relied on RNAi experiments. However, loss-of-function mutants allow for unbiased discovery of developmental phenotypes, as exemplified in our present study, in which we examined the role of Crc in later developmental stages, adult tissues and during aging.

Generally, misfolding-prone membrane proteins, such as Rh1^G69D, are thought to activate the PERK-mediated ISR response, among other ER stress responses (Donnelly et al., 2013). It is worth noting here that although both Drosophila and mouse models of adRP describe a protective role for Perk in retinal degeneration (Athanasiou et al., 2017; Chiang et al., 2012; Vasudevan et al., 2020), there has been conflicting evidence on the role of ATF4 in the mouse adRP model (Bhootada et al., 2016). In this study, we show that loss of crc accelerates the age-related retinal degeneration in a Drosophila model of adRP. As we have previously shown that Perk mutants similarly accelerate retinal degeneration in this model (Vasudevan et al., 2020), we interpret that crc mediates the effect of Perk in this model. Our data show that loss of Crc renders photoreceptors susceptible to age-related retinal degeneration in ninaE animals (solid red line, Fig. 4A). Along with our observation showing an increase in Crc protein levels in older flies (Fig. 4B,C), these data indicate that photoreceptors suffer from physiological stress which requires Crc for their proper function and survival during aging.

One of the visible phenotypes in adult crc mutants is ectopic wing venation (Fig. 5). It has previously been demonstrated that Gcn2 depletion in the posterior compartment of imaginal discs results in ectopic wing vein formation, although RNAi experiments from this study indicated this phenotype to be crc independent (Malzer et al., 2018). This raises the possibility of insufficient crc suppression in this previous study. The study proposed that GCN2 regulates bone morphogenetic protein (BMP) signaling by modulating mRNA translation in wing discs via eIF2α phosphorylation and Thor induction. Our results are consistent with this proposal, as Thor is a transcription target of crc. In addition, we report here that crc loss affects wing size, a finding that has not been reported previously. Given that BMP signaling has also been extensively implicated in determining wing size (Gibson and Perrimon, 2005; Shen and Dahmann, 2005), it is possible that GCN2-Crc signaling regulates wing size via BMP signaling. It is equally possible that GCN2-Crc signaling affects tissue size through regulating amino acid transport and metabolism through autonomous and non-autonomous means.

Although wing development is not known to be sexually dimorphic, fat tissues are known to have sex-specific effects, with particularly profound effects on female fertility in flies and other sexually dimorphic organisms (Valencak et al., 2017). It has been previously demonstrated that loss of crc in Drosophila larvae leads to reduced fat content and increased starvation susceptibility (Seo et al., 2009). Hence, it is possible that the block in oogenesis in crc mutants (Figs 6, 7) is due to metabolic changes in the female fat body, although this remains to be directly tested. This hypothesis integrates well with our data showing high Crc activity in adult fat tissues (Fig. S5A,B) and with observations from a previous study that amino acid sensing by GCN2 in Drosophila adult adipocytes regulates germ stem cells in the ovary (Armstrong et al., 2014). Indeed, our preliminary analysis with fat body-specific depletion of crc using the 3.1Lsp2-Gal4 driver leads to reduced fertility and increased dorsal appendage defects, similar to those seen in crc mutants (Fig. S4C). Nonetheless, it remains possible that Crc acts autonomously in the ovary but is undetectable using our current methods (Fig. S4B,C). In summary, our study has found utilities for the crc^GFSTF allele in discovering a new role for ISR signaling in disease models and during development, and also as an endogenous reporter for ISR activation.

Flies were reared on cornmeal-molasses medium at 25°C under standard conditions except for retinal degeneration experiments when they were reared under constant light. All fly genotypes and sources used in the study are listed in Table S1.

Full-length Drosophila eIF2α cDNA was amplified from DGRC plasmid (clone LD21861) with EcoRI and XbaI restriction sites using the following primers: Fwd, 5′-GGAATTCATGGCCCTGACGTCGCGCTTCTAC-3′; and Rev, 5′-GCTCTAGACTAATCCTCTTCCTCCTCCTCATCCTC-3′. The resulting DNA fragment was cloned into the EcoRI and XbaI sites of pUAST-attB to generate pUAST-attB-eIF2α^WT. For the eIF2α phosphorylation mutants, unique cut sites across the phosphorylation site were identified (AatII and AgeI), and the following gene fragments corresponding to S51A and S51D mutations were ordered from Integrated DNA Technologies (mutant residues underlined): eIF2α^S51A fragment, atggccctgacgtcgcgcttctacaacgagcggtatccggagatcgaggatgtcgttatggtgaacgtgctgtccatcgccgagatgggcgcctacgttcatctgcttgag tacaacaacatcgagggcatgatcctgctgtcggagctgGcccgccggcgcatccgctccatcaacaagctgattcgtgtcggcaagaccgaaccggtggtggtt; and eIF2α^S51D fragment, atggccctgacgtcgcgcttctacaacgagcggtatccggagatcgaggatgtcgttatggtgaacgtgctgtccatcgccgagatgggcgcctacgttcatctgcttgagtacaacaacatcgagggcatgatcctgctgtcggagctgGAccgccggcgcatccgctccatcaacaagctgattcgtgtcggcaagaccgaaccggtggtggtt. The pUAST-attB-eIF2α^WT plasmid described above was then restriction digested with AatII and AgeI, and the wild-type fragment was replaced with the synthetic mutant fragment (mutated nucleotides underlined) to generate pUAST-attB-eIF2α^S51A and pUAST-attB-eIF2α^S51D. The plasmids were then targeted to the VK13 attP-9A landing site [Bloomington Drosophila Stock Center (BDSC), 9732] by BestGene Inc to generate transgenic lines that were placed in the same 76A2 genomic locus.

Lethal phase analysis was performed as described previously (Vasudevan et al., 2020). Right wings were severed from 1- to 4-day-old flies and imaged using a Nikon SMZ1500 microscope outfitted with a Nikon 8MP camera with NIS-Elements software. Wing size was measured using the regions of interest feature in ImageJ software.

Female fertility was quantified by placing five 1- to 4-day-old virgin females with five yw males (or Dj-GFP males for fertilization assays) in a vial containing standard medium enhanced with yeast to encourage egg laying. After allowing 1 day for acclimatization, the flies were moved to a new vial and the number of eggs laid in a 24-h period were counted and quantified. Eggs were imaged for Fig. 6B by placing them on an apple juice plate and capturing them with a Nikon SMZ1500 microscope outfitted with 8MP Nikon camera controlled by NIS elements software. Ovaries from female flies in this experiment were dissected in ice-cold PBS and similarly imaged on apple juice plates for Fig. S4A.

Total RNA was prepared from five wandering third instar larvae using TriZol (Invitrogen), and cDNA was generated using random hexamers (Fisher Scientific) and Maxima H minus reverse transcriptase (Thermo Fisher Scientific) according to the manufacturer's protocol. qPCR was performed using PowerSYBR Green Mastermix (Thermo Fisher Scientific) using the following primers: crc: Fwd, 5′-GGAGTGGCTGTATGACGATAAC-3′, and Rev, 5′-CATCACTAAGCAACTGGAGAGAA5-3′; Thor: Fwd, 5′-TAAGATGTCCGCTTCACCCA-3′, and Rev, 5′-CGTAGATAAGTTTGGTGCCTCC-3′; and Rpl15: Fwd, 5′-AGGATGCACTTATGGCAAGC-3′, and Rev, 5′-CCGCAATCCAATACGAGTTC-3′.

Ovaries and fat bodies were dissected in ice-cold PBS from female flies reared for 2-3 days, along with yw males, on standard medium enhanced with dry yeast. Tissues were fixed in 4% paraformaldehyde (PFA) in PBS-Triton (0.2% Triton X-100 and 1× PBS) for 30 min, washed three times with PBS-Triton and blocked in PBS-Triton with 1% bovine serum albumin for 3 h (all at room temperature). Tissues were stained overnight at 4°C with the primary antibodies diluted in PBS-Triton, following which they were washed three times with PBS-Triton and incubated with Alexa Fluor-conjugated secondary antibodies (Invitrogen) in PBS-Triton for 3 h at room temperature. Tissues were mounted in 50% glycerol containing DAPI.

Eye imaginal discs were dissected from wandering third instar larva in ice-cold PBS and fixed in 4% PFA in PBS for 20 min, washed twice with PBS and permeabilized in 1× PBS-Triton for 20 min (all at room temperature). Discs were incubated in primary antibodies diluted in PBS-Triton for 2 h, washed three times in PBS-Triton, incubated in Alex aFluor-conjugated secondary antibodies (Invitrogen) in PBS-Triton for 1 h and washed three times in PBS-Triton, prior to mounting in 50% glycerol containing DAPI. Adult retinae were dissected and visualized with phalloidin as described previously (Huang et al., 2018). Antibodies used were as follows: phalloidin-Alexa 647 (1:1000, Invitrogen, A22287); chicken anti-GFP (1:500, Aves Labs, GFP-1020); rabbit anti-GFP (1:500, Invitrogen, A6455); rabbit anti-cleaved Dcp-1 (1:100, Cell Signaling Technology, 9578S); mouse anti-4C5 for Rh1 [1:500, Developmental Studies Hybridoma Bank (DSHB)]; rabbit anti-eIF2α (1:500, Abcam, ab5369); rabbit anti-S51 peIF2α (1:500, Abcam, ab32157); and rabbit anti-castor (1:50, gift from Dr Erika Bach, New York University Grossman School of Medicine). All images were obtained using a Zeiss LSM 700 confocal microscope with ZEN elements software and a 20× air or 40× water lens.

All experiments were performed in a white mutant background as crc^GFSTF, crc¹ and ninaE^G69D do not have eye color. Male flies 0-3 days old were placed (20 animals/vial) under 1000-lumen light intensity, and their pseudopupil structures (reflecting photoreceptor integrity) were monitored with a blue fluorescent lamp at 3-day intervals for a 30-day period. Medium was replaced every 3 days, and flies with disrupted pseudopupils in one or both eyes were marked as having retinal degeneration.

Fly head extracts were prepared from six severed male fly heads in 30 µl lysis buffer containing 10 mM Tris HCl (pH 7.5), 150 mM NaCl, protease inhibitor cocktail (Roche), 1 mM EDTA and 1% SDS. Following SDS-PAGE and western blotting, proteins were detected using primary antibodies and IRDye-conjugated secondary antibodies (LI-COR) [primary rabbit anti-GFP (1:500, Invitrogen) and mouse anti-beta Tubulin (1:1000, DSHB)] on an Odyssey system.

Female flies 0-3 days old were placed (10 animals/vial) in standard medium or in vials containing 5% sucrose and 2% agarose, prepared in distilled H₂O. The number of survivors was counted every 24 h and survivors were moved to new medium.

The quantified values used for graphs are listed in Table S2.

We thank Hugo Bellen's laboratory for making the crc MiMIC RMCE line available; Drs Lacy Barton and Lydia Grmai for discussions on the ovary phenotypes; and Drs Erika Bach and Jessica Treisman, and their laboratories, for helpful discussions that improved this work. We also thank the Bloomington Drosophila Stock Center [National Institutes of Health (NIH), P40OD018537] for supplying many of the fly stocks, and the Drosophila Genomics Resource Center (NIH 2P40OD010949) and FlyBase (U41 HG000739) for curating plasmids and sequence data used in this study.