Epidermal tyrosine catabolism is crucial for metabolic homeostasis and survival against high-protein diets in Drosophila

Metabolic adaptation to an altered nutritional environment is pivotal for animals. Because both shortage of and excessive nutrition can be detrimental, animals are equipped with various mechanisms to maintain metabolic homeostasis. Drosophila melanogaster, a model organism with evolutionarily conserved metabolic pathways and the capacity for efficient genetic manipulation, has provided valuable insights into the mechanisms of metabolic adaptation (Chatterjee and Perrimon, 2021). Among a myriad of nutrients, dietary protein, which serves as a precursor to amino acids (AAs), exerts a profound influence on animal physiology, including the growth, fertility and lifespan of fruit flies (Soultoukis and Partridge, 2016). Flies can modulate metabolism to maintain organismal homeostasis by detecting and adapting to alterations in dietary protein availability. Multiple organs are involved in AA perception. Primarily, sensory organs or neurons located in the gut rumen perceive the taste of AAs, which then transmit the signal to the central nervous system (Croset et al., 2016; Steck et al., 2018; Toshima et al., 2012; Yang et al., 2018). Subsequently, digested AAs can be directly sensed in the brain, eliciting neurotransmission (Bjordal et al., 2014; Manière et al., 2016), or triggering systemic adaptive responses in peripheral tissues, which often involves the secretion of hormones (Agrawal et al., 2016; Delanoue et al., 2016; Koyama and Mirth, 2016). The fat body represents a peripheral organ that plays a crucial role in AA sensing (Li et al., 2019). In response to a shortage of dietary AA intake, the deactivation of mechanistic target of rapamycin complex 1 (mTORC1) signalling in the fat body affects several endocrine hormones, including Stunted, Growth-blocking peptide 1 and 2, and Eiger (Manière et al., 2020). These fat body-derived soluble factors regulate the secretion of Drosophila insulin-like peptides (Dilps) from the brain that systemically stimulate insulin-like receptor (InR) to activate the insulin/IGF-1 signalling (IIS) pathway. IIS governs both growth and metabolism, partly through its downstream transcription factor FoxO, which targets various metabolic enzymes (Chatterjee and Perrimon, 2021; Link and Fernandez-Marcos, 2017). The fat body is believed to be a major metabolic organ, functioning as a counterpart of mammalian liver and white adipose tissue, in which many metabolic pathways are active. However, the contribution of other peripheral organs to the regulation of metabolic homeostasis in response to changing protein environments is poorly understood.

Nutritional sensing in the epidermal tissue is likely of great importance to insects because the presence of an exoskeleton limits the animal's final body size. Indeed, epidermal cells are highly responsive to nutrient-dependent endocrine signals, such as insulin receptor/phosphoinositide 3-kinase (PI3K) signalling, resulting in a marked alteration of cell size (Britton et al., 2002). Because the epithelium covers the entire body, it is fully exposed to haemolymph and possesses a great capacity to modulate circulating metabolites. Despite the long-standing study of the role of the epidermis in barrier function and cuticle formation, including sclerotisation and tanning during metamorphosis (Hopkins and Kramer, 1992), the extent to which the tissue responds and adapts to variations in dietary protein availability, particularly as a regulator of metabolism, remains to be elucidated.

Tyrosine (Tyr) is a non-essential amino acid required for catecholamine and melanin synthesis in addition to protein synthesis. Previously, we revealed that Tyr plays a crucial role in sensing dietary protein sufficiency (Kosakamoto et al., 2022). The scarcity of Tyr is perceived through activating transcription factor 4 (ATF4; Crc) in the fat body, which alters protein synthesis, mTORC1 signalling and feeding behaviour. To ensure enough Tyr levels, Drosophila larvae store massive amount of Tyr in the form of O-phospho-L-Tyr (Lunan and Mitchell, 1969; Mitchell and Lunan, 1964). In contrast, accumulation of Tyr can be detrimental due to its limited solubility and propensity to precipitate within cells as crystals (Scott, 2006; Wu, 2013). Hence, the metabolic homeostasis of Tyr during excessive protein consumption should be tightly regulated. Internal levels of Tyr can be modulated through its biosynthesis and degradation processes. Tyr is derived from phenylalanine (Phe) undergoing degradation into fumarate and acetoacetate for energy utilisation. Approximately 70% of Tyr is continuously degraded in human hepatocytes (Shiman and Gray, 1998). However, the contribution of the Tyr degradation pathway for metabolic adaptation and its regulatory mechanisms upon protein feeding remain unresolved.

In this study, we have unveiled the intricate regulation of enzymes responsible for the degradation of Tyr, which are conspicuously expressed in the epidermis. Their expression is highly responsive to dietary protein levels. The upregulation of Tyr catabolism is imperative for metabolic homeostasis and successful development under high-protein diets, underscoring the role of the epidermis as a metabolic regulator.

To investigate the metabolic adaptation of the epidermis to changes in the supply of dietary protein, we conducted RNA-sequencing analysis of the larval body wall, which comprises mainly the epidermis but also contains muscle, haemocytes, neurons and oenocytes. To manipulate the dietary protein composition, we fed third instar larvae diets consisting of yeast extract (YE) at varying concentrations: 4% (low), 10% (normal), and 24% (high). Feeding larvae these diets for 6 h led to changes in the gene expression profile in the body wall (Fig. 1A-D). Gene Ontology analysis revealed that the genes responsible for protein targeting from the endoplasmic reticulum to the Golgi apparatus and chaperones were induced by the high-protein diet (Fig. 1C). Additionally, Phe and Tyr catabolic processes were significantly upregulated (Fig. 1B,C, Table S1). These results suggest an acceleration of protein processing and AA degradation in response to the high load of dietary protein. Concurrently, the gene expression of AA transporters, such as sobremesa (sbm) or minidiscs (mnd) and Glutamine synthetase 2 was suppressed (Fig. 1D), indicating that the import and biosynthesis of AAs are downregulated to maintain AA homeostasis.

It was intriguing that Phe/Tyr catabolism was specifically induced when high-protein intake led to an increase in AA levels in general. We performed quantitative RT-PCR (qRT-PCR) for the genes encoding first and second enzymes in the Tyr degradation pathway, Tyrosine aminotransferase (Tat) and 4-Hydroxyphenylpyruvate dioxygenase (Hpd), respectively (Fig. 1B). We found a dose-dependent effect on the expression of both genes, although P-values were not less than the classical definition of statistical significance (P<0.05) for the comparison of 4% and 10% YE diets (Fig. 1E,F). These data suggest that the larval body wall can respond to changes in dietary protein levels and actively regulates Tyr metabolism.

The regulation of Hpd transcripts in response to changes in dietary protein levels motivated us to perform a detailed analysis of its molecular regulation. We utilised the CRISPR/Cas9 system to introduce monomeric ultrastable GFP (muGFP; Scott et al., 2018) at the C terminus of endogenous Hpd. As anticipated, increasing the YE concentration upregulated the GFP fluorescence in Hpd::muGFP larvae (Fig. 2A,B). Notably, the Hpd::muGFP signal was limited to the body wall and was not observed in visceral metabolic organs, such as fat body, gut, brain and wing disc (Fig. S1A-C). Furthermore, the GFP signal was observed in the dorsal epidermis but not in the ventral side (Fig. S1A,B). We also noticed GFP signals in oenocytes on the body wall (Fig. S1C). This body wall (carcass)-specific Hpd expression is consistent with the expression pattern demonstrated in the FlyAtlas Anatomy Microarray (Fig. S1D; https://flybase.org/reports/FBgn0036992). Not only Hpd but other Tyr catabolic enzymes are also highly expressed in the carcass, although Henna (Hn) is additionally expressed in the fat body (Fig. S1D, Flybase.org). Combining Hpd::muGFP with UAS-mCD8-RFP driven by the epidermis driver A58-Gal4, we discovered that Hpd reporter expression merges with epidermal cells but not with the muscle (stained by phalloidin) (Fig. 2C). Higher magnification images showed that Hpd::muGFP was localised to the cytoplasm and nucleus (Fig. 2D). Because the larval body wall has autofluorescence, we performed immunostaining using an anti-GFP antibody to check the pattern of Hpd::muGFP signals. The intensity of the GFP signal without staining correlated well with antibody staining (detected by Alexa Fluor 633), but the GFP antibody signals were enriched in the cytoplasm (Fig. S1E,F). In adult flies, we also observed an increase in Hpd::muGFP signals upon higher protein consumption (Fig. 2E). The signals were restricted to the epidermis of the head, thorax, abdomen, legs and oenocytes, and all these Hpd-expressing tissues responded well to an increase in dietary protein (Fig. S2A). In contrast, internal organs, such as gut, brain, ovary and Malpighian tubules, did not show any fluorescence even upon administration of a 24% YE diet (Fig. S2B).

As a second means of verifying that Hpd is expressed in the epidermis, we generated a GeneSwitch driver using the 500 bp upstream region of the Hpd start codon for drug-inducible gene manipulation in the epidermis. When the driver line was crossed with UAS-GFP, GFP was visible in the epidermis only when mifepristone (RU486), a drug that activates GeneSwitch, was added to the fly food (Fig. S3A,B). As expected, qRT-PCR revealed that Hpd expression in the epidermis was suppressed by Hpd knockdown using the driver with RU486 (Fig. S3C).

We next investigated how Hpd is induced by high-protein intake. As it is responsible for catalysing Tyr degradation, sensing the quantity of Tyr should be a key factor in the regulation of Hpd expression. We found that supplementation of a standard yeast-based diet with a high amount of Tyr alone induced Hpd::muGFP signal intensity, whereas the other nine non-essential AAs did not have an effect (Fig. 3A,B). We also found that Hpd::muGFP signal was increased when Phe, but not the other essential AAs, was added to the diet (Fig. 3A-C). Because Phe is a precursor of Tyr, we knocked down Henna (Hn), the enzyme responsible for converting Phe to Tyr, to distinguish the effect of Phe from the effect of Tyr. Hn-RNAi in the whole body suppressed the Hpd::muGFP upregulation induced by Phe supplementation (Fig. 3C), suggesting that increased Tyr, rather than Phe itself, is pivotal in the regulation of Hpd expression. We also found that the decreased Hpd expression induced by a low-protein diet was recovered by adding back Tyr (Fig. 3D). Furthermore, specific Tyr restriction for 3 days suppressed Hpd expression in adult flies (Fig. S4A,B). These data together suggest that dietary and/or internal Tyr levels modulate Hpd expression.

It is rational that an excess amount of a substrate should regulate the expression of the degradation enzyme; however, the molecular mechanism of such feedback regulation for Tyr degradation is not understood. Therefore, we explored how Hpd expression in the epidermis is regulated upon low- or high-protein diets. One candidate was ATF4, as the larval fat body senses Tyr scarcity via ATF4 (Kosakamoto et al., 2022). However, neither knockdown nor overexpression of ATF4 in the epidermis altered Hpd expression (Fig. S5A,B). In Caenorhabditis elegans, IIS induces Hpd protein expression via downregulation of DAF-16, an orthologue of FoxO (Lee et al., 2003). In rat hepatoma cells, the translation rate and the activity of Tat, the first enzyme in Tyr degradation, are increased by insulin treatment (Moore and Koontz, 1989). To test whether IIS is involved in Hpd regulatory mechanisms in Drosophila, we overexpressed the constitutively active form of InR in animals with Hpd::muGFP. The activation of IIS increased Hpd::muGFP signals upon a low- or standard-protein diet (YE 4% or 10%) (Fig. 3E,F). As expected, knockdown of InR by InR-RNAi in the epidermis suppressed Hpd::muGFP signals upon 10% YE diet (Fig. 3G), suggesting that basal Hpd expression is maintained by IIS. However, knockdown of InR or overexpression of the dominant-negative form of InR did not decrease Hpd expression upon a high-protein diet (Fig. 3G, Fig. S5B). This result suggests that IIS is not essential for mediating increased Hpd expression on a high-protein diet.

We next investigated whether FoxO is involved in Hpd regulation. Given that FoxO is negatively regulated by IIS, we assumed that lowered IIS in low-protein diets decreased Hpd expression via FoxO activation. Interestingly, however, downregulation of Hpd signals upon administration of the 4% YE diet was not affected by knocking down FoxO (Fig. 3E,F). Rather, increased Hpd signals upon normal- or high-protein diet (YE 10% or 24%) was suppressed by foxo-RNAi in the epidermis (Fig. 3E,F). Hpd transcript expression was also downregulated by foxo-RNAi during high-protein consumption (Fig. 3H). This outcome was unexpected because FoxO was believed to be inactivated by IIS in a high-protein context. To test whether FoxO is active in the epidermis of flies fed a high-protein diet, we performed qRT-PCR of the FoxO target gene Thor (also known as 4E-BP). Intriguingly, Thor expression showed a U-shaped curve: it was induced upon both low- and high-protein diets (Fig. 3I). These results together suggest that Hpd expression is induced by FoxO when larvae are fed a high-protein diet, theoretically independently of IIS. Because downregulation of Hpd expression with the 4% YE diet is independent of FoxO, there must be an additional mechanism that is dependent on decreased Tyr levels.

Our data suggested that high-protein stress activates FoxO by unknown mechanisms. To identify the upstream regulator of FoxO, we knocked down c-Jun N-terminal kinase (JNK; Bsk) or AMPK, two primary kinases known to activate FoxO (Brown and Webb, 2018). Knockdown of Drosophila JNK basket (bsk) or JNKK hemipterous (hep) did not suppress Hpd::muGFP signals (Fig. 3G). We confirmed this result by overexpressing the JNK inhibitor puckered (puc) or the dominant-negative form of bsk; these manipulations did not suppress the increased Hpd::muGFP signals induced by the high-protein diet (Fig. S5C,D). In contrast, AMPK knockdown downregulated the Hpd::muGFP signals caused by high-protein consumption (Fig. 4A-C). Increased Hpd::muGFP signals and Hpd gene expression in response to the high Tyr diet were both suppressed by AMPK knockdown (Fig. 4D,E). In this dietary condition, expression of the FoxO-target gene Thor was indeed downregulated (Fig. S6A), suggesting that AMPK activates FoxO and upregulates Hpd expression when the dietary Tyr level is high.

To analyse the AMPK activity, we performed western blotting of larval carcasses using an anti-phospho-AMPK antibody. We found that the p-AMPK/AMPK ratio was slightly but significantly increased by the high-protein diet and by the high Tyr diet (Fig. S6B-E). This increase was mainly attributed to a decrease of total AMPK, rather than an increase in p-AMPK (Fig. S6B-E). We noticed that this pattern of AMPK activation has also been reported in a previous study (Stenesen et al., 2013) wherein the increase of AMPK activity in a long-lived mutant for AMP biosynthesis was concluded to be due to the decreased total AMPK. Thus, our data suggest that AMPK in the epidermis could be activated by increased protein or Tyr consumption.

It is believed that AMPK can recognise energy shortage by sensing the AMP/ATP ratio (Hardie, 2014). As expected, the whole-body AMP level was increased upon administration of the high-protein diet (Fig. S6F). However, AMP/ATP ratio was unchanged as the ATP level was also increased by the high-protein diet, at least in the larval whole body. This result suggests that AMPK is activated independently of the AMP/ATP ratio in response to a high concentration of dietary protein. In mammalian AMPK complexes, three distinct gamma subunits (binding to adenine nucleotides) respond differently to AMP, ADP and ATP (Ross et al., 2016). Given that the AMP/ATP ratio is not necessarily an exclusive determinant of AMPK activity, we cannot rule out the possibility that AMP levels predominantly regulate the Drosophila AMPK activity and are less influenced by ATP. Alternatively, the AMP/ATP ratio might be specifically downregulated in the epidermis, but not in the whole body. Detailed biochemical analysis will be necessary to determine how AMPK is regulated by high-protein or high-Tyr diets.

To elucidate the physiological meaning of Hpd induction, we examined whether the Hpd and Tyr degradation pathways contribute to the metabolic homeostasis and survival of animals. Notably, whereas control larvae exhibited the ability to maintain internal Tyr levels despite being fed a high-protein diet, Hpd knockdown in the epidermis led to a massive increase in overall Tyr levels only under this dietary condition (Fig. 5A). We observed an increase in the downstream metabolite dopamine in the larvae with Hpd-RNAi, but levels of the precursor L-DOPA remained unchanged (Fig. 5B,C). Knockdown of foxo in the epidermis similarly resulted in elevated Tyr levels (Fig. 5D). Therefore, FoxO-mediated Hpd induction is necessary for larvae to maintain Tyr metabolism during protein overfeeding.

As anticipated, Hpd knockdown using epidermis driver, Hpd mutation or foxo mutation all significantly impaired development when on high-protein diets, leading to exacerbated lethality at both larval and pupal stages (Fig. 5E-J). By contrast, Hpd knockdown using other Gal4 drivers (elav-Gal4 for the brain, r⁴-Gal4 for the fat body, NP1093-Gal4 for the Malpighian tubules, and NP1-Gal4 for the gut) did not increase lethality (Fig. S7A-H). This result supports our conclusion that the epidermis plays a crucial role in maintaining Tyr homeostasis. In Drosophila larvae, Tyr is known to be stored, for tanning the pupal case, as a phosphate (O-phospho-L-Tyr) to enhance its solubility (Lunan and Mitchell, 1969; Mitchell and Lunan, 1964). Considering that this metabolite is important for the development, it was possible that altered levels of O-phospho-L-Tyr could be a cause of lethality upon knockdown of Hpd. High-protein diet led to a decrease in the level of O-phospho-L-Tyr, possibly due to developmental delay. Knockdown of Hpd increased the O-phospho-L-Tyr levels, but it did not reach the level induced by the 10% YE diet (Fig. 5K), suggesting that this metabolite seemed is not responsible for the lethality. The exact cause of lethality was not known, but we surmise that impairment of Tyr metabolism for catecholamines and/or accumulation of excess Tyr would be toxic for animals. Taken together, our data emphasise the crucial role of Tyr catabolism in the epidermis for successful development under high-protein conditions (Fig. 6).

In this study, we demonstrate that the Tyr degradation pathway acts in the epidermis, which is highly responsive to varying dietary protein levels by sensing Tyr levels. Using a GFP-fused Hpd reporter line, we reveal that low-protein diets decrease Hpd expression via reduced IIS activity. By contrast, Tyr degradation is upregulated by a high-protein diet through the AMPK-FoxO axis. The tight regulation of Tyr metabolism in epidermal cells is pivotal for successful development under high-protein stress in Drosophila larvae. Therefore, we propose that epidermis is a nutrient-responsive organ that maintains organismal AA homeostasis, especially for Tyr.

The Tyr degradation pathway has been shown to be crucial for the survival of hematophagous animals, which experience an increase in hazardous Tyr levels after blood digestion that must be mitigated through degradation (Sterkel et al., 2016). Hpd-silenced animals exhibit Tyr precipitation in the gut after a blood meal, which can result in lethality (Sterkel et al., 2016). Given that blood-sucking animals frequently transmit infectious diseases, vector control is of utmost importance, although the precise mechanism by which Tyr degradation enzymes are regulated in vivo is not yet entirely understood. Our present study reveals that the AMPK–FoxO axis regulates extensive Tyr degradation in the insect epidermis, providing valuable insights for efficient vector control in the future. Notably, Hpd inhibition did not kill Drosophila larvae fed a normal diet. Therefore, selective vector control, which is less damaging to other insects and ecosystems, could be achieved via the regulation of Tyr catabolism.

A previous study reported that the Tyr degradation pathway is related to lifespan regulation (Parkhitko et al., 2020). The extension of lifespan in long-lived flies is attributed to the downregulation of Tyr degradation and an increase in dopamine levels. In C. elegans, Hpd is an important target of DAF-16 (a FoxO orthologue) for lifespan extension in the daf-2 (an orthologue of InR) mutant (Lee et al., 2003). The DAF-16 and FoxO binding sites are conserved in the upstream region of the Hpd gene in both C. elegans and D. melanogaster (Lee et al., 2003). Our data showed that FoxO upregulates Hpd during high-protein feeding, whereas Lee et al. found that DAF-16 downregulates Hpd. This contradiction may be due to different nutritional conditions. In normal dietary conditions, FoxO would be the suppressor of Hpd, but when excess protein is ingested, the function of FoxO is reversed. We found that AMPK is responsible for the FoxO-mediated Hpd induction during high-protein or high-Tyr stress. It has been reported that the elevation of internal Tyr levels by a genetic mutation in the Tat gene in C. elegans activates AMPK by unknown mechanisms (Ferguson et al., 2013). Furthermore, AMPK could be activated by β-hydroxybutyrate derived from Tyr catabolism (Tong et al., 2021). These findings suggest that feed-forward regulation increases Hpd expression when Tyr levels are elevated. Given that our data suggest that Tyr specifically regulates Hpd expression among AAs, it is conceivable that AMPK is directly regulated by Tyr or its metabolite levels. A comprehensive examination of the mechanisms involved in the AMPK–FoxO regulation in response to high-protein stress is further needed.

In humans, Tyr catabolism is primarily regulated in the liver and to a lesser extent in the kidney (Noda and Ichihara, 1976). It has been reported that people with tyrosinemia, a condition marked by a deficiency in the enzymes involved in Tyr degradation, exhibit tissue damage and disruption of hepatic and neurological functions due to crystal formation of Tyr and accumulation of intermediate metabolites from the Tyr degradation pathway (Adnan and Puranik, 2022; Najafi et al., 2018; Xie et al., 2019). To mitigate symptoms, treatment is given to block the formation of toxic Tyr metabolites and individuals must decrease their dietary intake of Phe and Tyr (Adnan and Puranik, 2022), emphasising the evolutionarily conserved significance of proper Tyr metabolism in humans. Recent studies identified Tyr as a predictive risk factor for the progression of diabetes and obesity (Newgard et al., 2009; Stančáková et al., 2012; Wang et al., 2011; Würtz et al., 2012). Additionally, decreased expression of Tat and Hpd in the liver is associated with poor prognosis for cancer, highlighting the role of Tyr metabolism in tumour suppression (Fu et al., 2010; Tong et al., 2021; Yang et al., 2020). Therefore, a more profound comprehension of the regulatory mechanisms governing Tyr metabolism has the potential to inform effective clinical interventions for such diseases.

The Drosophila counterparts of the mammalian liver are the fat body and oenocyte (Gutierrez et al., 2007; Li et al., 2019). Although the fat body serves as a primary metabolic organ, it does not apparently participate in Tyr degradation. Although we found that Tyr degradation primarily occurs in the epidermis, our previous research has shown that the fat body is an essential organ for sensing Tyr in order to adapt to the dietary protein environment (Kosakamoto et al., 2022). Thus, inter-organ communication likely exists between the epidermis, which maintains internal Tyr levels, and the fat body, which monitors Tyr levels. The specialised role of the epidermis in Tyr degradation may be attributed to the organ's significant demand for Tyr for pigmentation and sclerotisation during metamorphosis (Gorman and Arakane, 2010). To prepare for the metamorphosis, large amounts of Tyr are stored in the haemolymph as a form of phosphate in Drosophila larvae (Lunan and Mitchell, 1969; Mitchell and Lunan, 1964). Similarly, in Calpodes larvae, Tyr is stored in vacuoles within the fat body and released into the haemolymph during metamorphosis (McDermid and Locke, 1983). The transportation of large amounts of Tyr to the epidermis and acute acceleration of melanogenesis carries a risk of oxidative damage. Therefore, degradation machinery for Tyr is likely to be necessary in the epidermis to prevent damage. In larvae fed a high-protein diet, it is reasonable to use the innate epidermal Tyr-degradation system to efficiently counteract the toxicity, rather than ectopically activating it in the fat body or other tissues. Nonetheless, even after completing metamorphosis, adult flies express Hpd in epidermal tissue as well as in oenocytes. The evolutionary specialisation of the epidermis in Tyr degradation in insects is a fascinating topic. The unexpected participation of the epidermis in animal metabolic homeostasis broadens our perspective of the adaptation mechanism to a changing nutritional environment.

Flies were reared on a standard yeast-based diet containing 4.5% cornmeal (NIPPN CORPORATION), 6% brewer's yeast (Asahi Breweries, HB-P02), 6% glucose (Nihon Shokuhin Kako) and 0.8% agar (Ina Food Industry, S-6) with 0.4% propionic acid (Wako, 163-04726) and 0.15% butyl p-hydroxybenzoate (Wako, 028-03685). Flies were maintained under conditions of 25°C. To allow synchronised development and constant density, embryos were collected onto agar plates (2.3% agar, 1% sucrose and 0.35% acetic acid) with live yeast paste.

The fly lines used in this study were w^Dah (Grandison et al., 2009), foxo^Δ94 (Slack et al., 2011), A58-Gal4 (Galko and Krasnow, 2004), elav-Gal4 [Bloomington Drosophila Stock Center (BDSC), 458], r⁴-Gal4 (BDSC, 33832), NP1093-Gal4 (Kyoto Drosophila Stock Center, 103880), NP1-Gal4 (Kosakamoto et al., 2020), da-Gal4 (Kosakamoto et al., 2022), UAS-mCD8-RFP (BDSC, 27391), UAS-GFP (BDSC, 6874), UAS-ATF4 (FlyORF, F000106), UAS-Hpd-RNAi [Vienna Drosophila Resource Center (VDRC), 103482], UAS-foxo-RNAi (BDSC, 32429), ATF4-RNAi (VDRC, 109014), UAS-lacZ-RNAi (from Dr R. Carthew, Northwestern University, IL, USA), UAS-InR^DN (BDSC, 8253), UAS-InR^CA (BDSC, 8263), UAS-bsk^DN (Kyoto Drosophila Stock Center, 108773), UAS-puc (from Dr Igaki, Kyoto University, Japan), UAS-AMPK-RNAi (BDSC, 57785), UAS-InR-RNAi (BDSC, 31594), UAS-bsk-RNAi (National Institute of Genetics, HMS00777), UAS-hep-RNAi (National Institute of Genetics, 4353R-2), UAS-Hn-RNAi (VDRC, 1100511).

To generate Hpd::muGFP flies, we used the CRISPR/Cas9 system to insert monomeric ultrastable GFP (muGFP) at the C terminus of the Hpd gene (Kina et al., 2019). The codons of muGFP were optimised for expression in Drosophila melanogaster. These modifications and insertions into the EcoRI/XbaI site in the pUC57 vector were performed by GenScript. An sgRNA target site of Hpd was selected using CRISPR Optimal Target Finder (Gratz et al., 2014). Complementary oligonucleotides with overhangs were annealed and cloned into the BbsI-digested U6b vector using a DNA ligation kit (Takara Bio, 6023): sense strand, TTCGGAAATTGAACAAGCCAAGCG; antisense strand, AAACCGCTTGGCTTGTTCAATTTC.

The targeting vector was constructed by inserting muGFP with 500 bp homology arms of the Hpd locus with a mutation in the PAM sequence next to the sgRNA site (from AGG to AGA). First, muGFP and homology arms were PCR-amplified using Q5 High-Fidelity 2× Master Mix (New England Biolabs, M0492L). Primers for PCR were designed using the NEBuilder Assembly Tool. The gel-purified PCR products were cloned into the EcoRI-digested pBluescript II SK(+) vector using NEBuilder HiFi DNA Assembly Master Mix (New England Biolabs, E2621X). The mixture of pU6b-sgRNA and targeting vector was microinjected into w1118; attP40{nos-Cas9}/CyO embryos by WellGenetics. The F0 adults were crossed with balancer lines, and the muGFP-inserted lines were selected by PCR amplification of the target locus. The primers used for PCR are listed in Table S2.

To generate pHpd-GeneSwitch flies, 500 bp upstream of the start codon of Hpd and the GeneSwitch sequence were PCR-amplified using Q5 High-Fidelity 2× Master Mix. Primers for the PCR were designed using the NEBuilder Assembly Tool. The gel-purified PCR products were cloned into the KpnI-digested pElav-GeneSwitch vector using NEBuilder HiFi DNA Assembly Master Mix. The plasmid was injected into w1118 embryos by WellGenetics. The F0 adults were crossed with balancer lines, and w+ lines were selected. The primers used for PCR are listed in Table S2.

Embryos were harvested under the standard (6%) yeast-based diet until they became third instar larvae. Then, the larvae were floated using 30% glycerol and transferred using a soft brush to fly vials containing various diets and incubated for 6-24 h. The sampling stage of larvae was set to 92-96 h after egg laying. For protein manipulation, 4%, 10% or 24% yeast extract (Nacalai Tesque, 15838-45) was mixed with 6% glucose, 1% agar, 0.3% propionic acid and 0.15% nipagin (YE diet). For the survival assay, we added 1% baker's yeast (LeSaffre, Saf-Instant Red yeast) to the YE diet above to promote larval development.

For RNA-sequencing analysis of larval carcasses, the larvae were fed three types of diets for 6 h from 90 h after egg laying. Total RNA was purified from six carcasses of female third instar larvae using a Promega ReliaPrep RNA Tissue Miniprep kit (z6112). Triplicate samples were prepared for each experimental group. RNA was sent to Kazusa Genome Technologies for 3′ RNA-sequencing analysis. The cDNA library was prepared using the QuantSeq 3′ mRNA-Seq Library Prep Kit for Illumina (FWD) (Lexogen, 015.384). Sequencing was performed using Illumina NextSeq 500 and NextSeq 500/550 High Output Kit v2.5 (75 cycles) (Illumina, 20024906). Raw reads were analysed using the BlueBee Platform (Lexogen) for trimming, alignment to the Drosophila genome, and counting of the reads. Count data were analysed by the Wald test using DESeq2.

For qRT-PCR analysis, total RNA was purified from the whole body or dissected carcass from third instar larvae using a Promega ReliaPrep RNA Tissue Miniprep kit (z6112). The cDNA was synthesised from 400 ng of DNase-treated total RNA using the Takara PrimeScript RT Reagent Kit with gDNA Eraser (Takara Bio, RR047B). qRT-PCR was performed using TB Green™ Premix Ex Taq™ (Tli RNaseH Plus) (Takara Bio, RR820W) and a QuantStudio 6 Flex Real Time PCR system (Thermo Fisher Scientific) using RpL32 as an internal control. Primer sequences are listed in Table S2.

Metabolites were measured by ultra-performance liquid chromatography-tandem mass spectrometry (LC‒MS-8050/LCMS-8060, Shimadzu) based on the Package for Primary Metabolites Ver. 2 (Shimadzu) (Kosakamoto et al., 2022; Shiota et al., 2018). Five whole bodies of third instar larvae were homogenised in 160 μl of 80% methanol containing 10 μM internal standards (methionine sulfone and 2-morpholinoethanesulfonic acid). After centrifugation at 14,000 g at 4°C for 5 min, 150 μl of supernatant was mixed with 75 μl of acetonitrile and deproteinised. After centrifugation at 20,000 g at 4°C for 5 min, the supernatant was applied into a prewashed 10-kDa centrifugal device (Pall, OD010C35), and the flow-through was completely evaporated using a centrifugal concentrator (TOMY, CC-105). The samples were resolubilised in ultrapure water and injected into the LC‒MS/MS with a PFPP column (Discovery HS F5 (2.1 mm×150 mm, 3 μm; Sigma-Aldrich) in the column oven at 40°C. A gradient of solvent A (0.1% formic acid, water) to solvent B (0.1% formic acid, acetonitrile) for 20 min was used to separate solutes. The MRM parameters were optimised by the injection of the standard solution through peak integration and parameter optimisation with the LabSolutions LCMS software (LabSolutions, Shimadzu). For ATP measurement, resolubilised samples were diluted fivefold and subjected to the luminometric ATP assay according to the manufacturer's protocol (Dojindo, 346-09793).

To analyse Hpd::muGFP expression in larvae, the animals were picked up and left in embryo dishes filled with Milli-Q water for 8 min at −30°C. When larval movement was restricted, images were acquired using a fluorescence stereomicroscope (Leica, MZ10F). GFP fluorescence was quantified using Fiji by calculating the average intensity of three regions of interest per body segment of each larva.

For more detailed analysis of the expression pattern of Hpd, the larval epidermis was dissected as described previously (Ramachandran and Budnik, 2010). Briefly, a larva was picked up and placed in a drop of PBS and then pinned at the anterior and posterior ends with two pins. To observe the dorsal side, the ventral side was placed at the top. Using microscissors, the larva was cut from the posterior side to the anterior end. After removing the internal organs, the epidermis was gently stretched and pinned by four corner pins. The sample was fixed by replacing PBS with 4% paraformaldehyde in PBS. After fixation for 20 min at room temperature (RT), the sample was washed multiple times with PBST (PBS with 0.1% Triton X-100) and incubated for 2 h at RT with Hoechst 33342 (Invitrogen, H3570; diluted to 0.4 mM, 1:100) and Phalloidin-iFluor 647 (Abcam, ab176759; 1:5000). For comparison of raw GFP and antibody-stained GFP, the fixed epidermis was incubated with anti-GFP antibody (Nacalai Tesque, 04404-26; 1:200) overnight at 4°C. After three washes with PBST, the sample was incubated for 2 h at RT with anti-rat Alexa Fluor 633 secondary antibody (Invitrogen, A-21094: 1:200), Hoechst 33342 (Invitrogen, H3570; diluted to 0.4 mM, 1:100) and Phalloidin-iFluor 555 (Abcam, ab176756; 1:5000). After washing, tissues were mounted in 80% glycerol and observed using a Leica TCS SP8 confocal microscope.

The carcasses of five larvae were dissected in PBS and homogenised in 50 μl of RIPA buffer (Fujifilm Wako, 188-02453) supplemented with protease inhibitor (Fujifilm Wako, 165-26021) and phosphatase inhibitor cocktails (Roche, 4906845001). The supernatant was collected after centrifugation (15,000 g for 5 min at 4°C) and the protein amount was quantified by BCA assay (Fujifilm Wako, 164-25935). The samples were mixed with 6× SDS-PAGE sample buffer (Nacalai Tesque, 09499-14) and 10 μg protein samples were subjected to standard SDS-PAGE. Gels were transferred to the PVDF membrane and blocked with EveryBlot Blocking Buffer (Bio-Rad, 12010020). Primary antibodies used in the study were anti-tubulin [DM1A +DM1B] (Abcam, ab44928; 1:1000), anti-histone H3 (Cell Signaling Technology, 1B1B2; 1:1000), anti-phospho-AMPKα (Cell Signaling Technology, 2535S; 1:500), anti-total AMPKα (Cell Signaling Technology, 2532S; 1:200). Horseradish peroxidase (HRP)-conjugated secondary antibodies were: anti-mouse IgG, HRP-linked antibody (Cell Signaling Technology, 7076S; 1:1000) and anti-rabbit IgG, HRP-linked antibody (Cell Signaling Technology, 7074S; 1:1000). The signals were visualised by chemiluminescence using Immobilon (Millipore, WBLUF0100) and detected using an Amersham ImageQuant 800 imager (Cytiva).

Statistical analysis was performed using GraphPad Prism 8 or 9. The sample numbers were determined empirically. All data points were biological, not technical, replicates. No data were excluded. An unpaired and two-sided Student's t-test was used to test between samples. One-way ANOVA with Holm–Šídák's multiple comparison test was used to test among groups. One-way ANOVA with Dunnett's multiple comparison test was used to compare with a control sample. All experimental results were repeated at least twice to confirm the reproducibility. Bar graphs are drawn as the mean and s.e.m.

We would like to acknowledge Kyoto Stock Center, National Institute of Genetics, Vienna Drosophila Resource Center, and Bloomington Drosophila Stock Center for reagents. We thank the RIKEN Kobe BioImaging Facilities & Factory for microscopy, and the DNA Analysis Facility, Laboratory for Developmental Genome System for RNA-sequencing analysis. We thank all members of our laboratory for technical assistance and critical advice.

The peer review history is available online at https://journals.biologists.com/dev/lookup/doi/10.1242/dev.202372.reviewer-comments.pdf