Nutrition-dependent juvenile hormone sensitivity promotes flight-muscle degeneration during the aphid dispersal-reproduction transition

Dispersal is an evolutionary driver that fundamentally shapes the distribution, abundance and diversity of insects, and plays a key role in their persistence in heterogeneous habitats (Weigang and Kisdi, 2015). However, dispersal is associated with high energetic costs (Bonte et al., 2012), which require trade-offs with other physiological activities such as reproduction (Guo et al., 2016). After locating a suitable colonization environment, insects that have dispersed are able to integrate external and internal environmental signals and maintain a systemic energy balance, while shifting resources toward the production of offspring instead of flight (Zera and Denno, 1997; Zhang et al., 2019).

A life-history trade-off exists between flight capability and reproduction in many wing-dimorphic insects (Guerra, 2011). Aphids can develop either with or without wings. Unwinged aphids remain on the host plant and feed and reproduce there. Winged aphids are induced by environmental cues, such as crowding and reduced nutritional quality. When a winged aphid molts into the adult form, the aphid does not feed before flight. After the aphid flies to a new host plant and begins feeding, the flight muscles (which represent a valuable nutritional resource) degenerate, and presumably the freed resources are redirected to the development of embryos (Brisson, 2010; Johnson, 1959; Kobayashi and Ishikawa, 1993). Thus, muscle degeneration is an adaptive strategy to maximize reproductive success. It was previously shown that topical application of juvenile hormone (JH) induced flight-muscle degeneration, which resembled feeding-dependent degeneration (Kobayashi and Ishikawa, 1994). Nevertheless, the regulatory mechanisms underlying flight-muscle degeneration in aphids when flight is unnecessary remains poorly understood.

Hormonal signals commonly integrate environmental cues (such as nutrition) to mediate a range of developmental processes (Nijhout, 1999; Beldade et al., 2011; Simpson et al., 2011; Johnson et al., 2014; Zhang et al., 2019). Besides regulating metamorphosis and reproduction (Jindra et al., 2013; Roy et al., 2018), JH plays key roles in regulating developmental plasticity in insects, including dispersal-reproduction trade-offs (Gotoh et al., 2014; Reiff et al., 2015; Miura, 2018). Target of rapamycin complex 1 (TORC1) is a well-conserved nutrient sensor (mainly amino acids) in the insect fat body, with a prominent role in linking systemic nutritional conditions to local cellular physiology in peripheral tissues (Colombani et al., 2003; Géminard et al., 2009; Howell and Manning, 2011; Li et al., 2019b). Eukaryotic initiation factor 4E-binding protein (4EBP) is a key TORC1 target, and the TORC1-mediated phosphorylation of 4EBP is regarded as an indicator of TORC1 activity (Hietakangas and Cohen, 2009). Recent studies have also implicated the TORC1 pathway in the stimulation of JH biosynthesis via upregulation of the expression of juvenile hormone acid methyltransferase (JHAMT) (Maestro et al., 2009; Pérez-Hedo et al., 2013; Lu et al., 2016; Zhu et al., 2020), which encodes a rate-limiting enzyme for JH biosynthesis and is expressed specifically in the corpora allata located in insect heads (Shinoda and Itoyama, 2003; Wen et al., 2015). In addition to systemic biosynthesis, gene expression levels of other molecules required for JH signaling, such as the JH receptor Met (Konopova and Jindra, 2007), also affect JH regulatory efficacy. JH/Met induces the expression of the JH primary-response gene Kr-h1, which conservatively reflects the degree of JH signaling (Jindra et al., 2013; Li et al., 2019a).

To examine how nutrition and JH interact to regulate flight-muscle degeneration, we focused on the winged female morph of pea aphids. We report that TORC1 mediates nutritional signaling and enhances the expression of Met and Kr-h1, and thus JH sensitivity, specifically in the flight muscles, leading to the restriction of energy supplies and the induction of flight-muscle degeneration in winged aphids. Our results highlight hormone sensitivity as an important tissue-specific regulatory mechanism of developmental transitions in the insect life history.

Our cage studies indicated that newly molted, winged female aphids remained static for 24 h (resting mode). The females then flew toward the light and maintained an active state within a 16 h photoperiod (flight mode). Females that successfully located plants for colonization then initiated feeding (settling mode), which resulted in fully distended abdomens by 48-60 h (colonization mode). By 72-84 h post eclosion, nearly all females began to produce nymphs (larviposition mode) (Fig. 1A-A″).

Examining the flight muscles by transmission electron microscopy (TEM) showed that 24-h-old females in flight mode possessed large, irregularly shaped mitochondria that were highly abundant between muscle fibers (Fig. 1B-D; Fig. S1A). The flight muscles of females in colonization mode (48 h) contained irregularly shaped mitochondria with condensed matrices, but exhibited no obvious defects in the muscle fibers (Fig. 1B′,C′,D; Fig. S1B). Subsequently, muscle fibers were severely degenerated with few mitochondria present in larviposition-mode females (72 h) (Fig. 1B″,C″,D; Fig. S1C). Until 96 h after eclosion, almost all the flight muscles were degenerated in feeding aphids (Fig. 1B‴,C‴,D; Fig. S1D). In contrast, when winged females were starved after eclosion, the flight-muscle fibers remained morphologically intact over a 96 h observation period, although mitochondria shifted from a flattened to a circular morphology (Fig. 1E-F‴; Fig. S2A-D). We further noted that in winged female aphids, muscle degeneration induced by feeding was present in the flight muscles but not the leg muscles, in which JH signaling (as indicated by gene expression levels of Met and Kr-h1) slightly decreased in feeding aphids compared with that in starved aphids (Fig. S3A,B).

We then examined whether exogenous JH plays a role similar to feeding in inducing morphological changes in the flight muscles during starvation. Topical application of methoprene (a JH mimic) significantly induced Kr-h1 expression in the flight muscles at 6-48 h after treatment during starvation (Fig. 2A). Methoprene application also induced flight-muscle degeneration; similarly, mitochondrial disruption occurred earlier than muscle-fiber breakdown (Fig. 2B-F; Fig. S4A-D). Although methoprene induced Kr-h1 expression in the leg muscles, the induction efficiencies were significantly lower than those in the flight muscles at 12 h after application (Fig. S5A). Notably, methoprene application did not significantly induce leg-muscle degeneration (Fig. S5B). It is likely that feeding activates tissue-specific JH signaling and induces flight-muscle degeneration in winged female pea aphids.

We compared the transcriptional dynamics of genes with roles in JH biosynthesis and signaling between winged morph aphids that were fed or starved. We firstly employed RNA-sequencing (RNA-seq) analysis of the heads to compare the transcriptional changes of all genes involved in JH biosynthesis between winged aphids in ‘flight’ (24 h post eclosion) and ‘colonization’ (60 h post eclosion) modes. The results showed that there were no significant changes in expression in any of the JH-biosynthesis genes in the heads between the two modes (Fig. S6A). The pea aphid might have four JHAMT homologs; however, there is no clear evidence showing which gene(s) might be the functional JHAMT homolog(s). JHAMT1, JHAMT2 and JHAMT3 were annotated in the transcriptome, and we further verified their expression levels using quantitative real-time PCR (qPCR) under the starvation versus feeding regime. The relative transcript abundance of JHAMT1-3 in the heads of fed and starved winged females was similar up to 72 h post eclosion (flight-muscle degeneration had already initiated by this time), but diverged post 72 h. The transcript abundance of JHAMT2 increased in feeding aphids, whereas that of JHAMT3 decreased in starved aphids, and significant differences in JHAMT expression were observed until 96 h post eclosion between feeding and starved females (Fig. 2G-I). Conversely, the relative transcript abundance of Met and Kr-h1 was significantly higher in the thoraxes (which are almost filled with flight muscles) of fed versus starved females by 48-96 h post eclosion (Fig. 2J,K).

The temporal expression patterns of key genes of the JH pathway implied that the upregulated expression of the JH intracellular receptor gene Met resulted in the increase of JH signaling, even though JH-biosynthesis genes were not significantly upregulated during the colonization of winged aphids. Given these outcomes, we further explored the spatial expression of Met and Kr-h1. We compared Met and Kr-h1 expression in the heads, thoraxes, ovaries, fat body and legs with that in the flight muscles of winged females at 60 h post eclosion under starved or fed conditions. The results showed that each gene was preferentially expressed in the thoraxes and flight muscles of the fed winged aphids, especially Met (Fig. 2L,M).

Temporal and spatial expression patterns of JH biosynthesis and signaling genes demonstrated that the transcription of Met was specifically activated in the flight muscles of winged aphids during the process of colonization. It is reasonable to assume that many intracellular receptors contribute to the sensitive detection of JH levels and subsequently induce JH nuclear signals. To verify this hypothesis, we topically applied the JH mimic on the winged aphids during colonization mode and detected the expression levels of Kr-h1 in various tissues using qPCR 12 h after the treatment. The results showed that JH induced the expression of Kr-h1 more efficiently in the flight muscles than in any other tissues of feeding winged aphids (Fig. 2N).

Subsequently, we employed RNA interference (RNAi) to decrease the feeding-induced gene expression of Met and Kr-h1. Met RNAi was not successfully achieved in the pea aphid, whereas Kr-h1 RNAi resulted in ∼50% reduction in gene expression in the flight muscles of feeding winged aphids 48 h after the treatment (Fig. 3A). Importantly, aphids treated with Kr-h1 RNAi exhibited delayed mitochondrial disruption and muscle-fiber breakdown compared with control aphids (Fig. 3B-D; Fig. S7A,B).

As shown above, JH sensitivity-induced flight-muscle degeneration depends on feeding (nutrition). The TORC1 pathway couples nutrition with JH biosynthesis in many insects (Maestro et al., 2009; Pérez-Hedo et al., 2013; Lu et al., 2016; Zhu et al., 2020). We then assessed whether TORC1 activity was also associated with nutrition-dependent JH sensitivity and flight-muscle degeneration in the process of the flight-reproduction life-history transition in aphids.

Limited by the efficiency of RNAi treatment, we used the TOR protein inhibitor rapamycin (Edwards and Wandless, 2007). About 60% of rapamycin-treated aphids died intensively by 72 h after the treatment (Fig. 3E). As expected, rapamycin reduced TORC1 activity in the whole body, excluding embryos, as evidenced by reduced 4EBP phosphorylation at 36, 60 and 72 h after the treatment (Fig. 3F). Rapamycin treatment specifically decreased the expression of Met and Kr-h1 in the flight muscles, but had no effect on JHAMT1-3 expression in the heads at 36 h after the treatment (Fig. 3G). By 60 h after rapamycin treatment, the effects of the TORC1 pathway could be observed in the regulation of JHAMT expression. The expression levels of JHAMT2 in the heads of rapamycin-treated aphids were significantly downregulated at this time point (Fig. 3H), by which most flight muscles of the control aphids, which were already in larviposition mode, had degenerated. TEM analysis further showed that rapamycin treatment significantly inhibited flight-muscle degeneration, whereas methoprene application to rapamycin-treated aphids rescued Kr-h1 expression and flight-muscle degeneration by 48 h after the treatment (Fig. 3I-M; Fig. S8A-C). Thus, TORC1-mediated nutritional signaling promotes JH sensitivity and signaling in the flight muscles, and causes muscle degeneration.

We conducted two parallel RNA-seq experiments to further explore the molecular mechanism of JH-induced flight-muscle degeneration. In the two groups of experiments, JH signals were activated either by feeding or by artificial JH application in starvation (Fig. 4A,D). Principal component analysis (PCA) results showed high consistency between biological replicates of each treatment (Fig. S9A). RNA-seq analysis of the flight muscles identified 5160 differentially expressed genes between ‘flight’- and ‘colonization’-mode females (Fig. 4B); the top pathways enriched in the KEGG aggregation analysis are listed in Table S4. A number of genes that are involved in the proteasome, protein processing and export, autophagy and the phagosome pathway were significantly upregulated, whereas other genes that regulate the citrate cycle (tricarboxylic acid or TCA cycle) of 2-oxocarboxylic acid or carbohydrate metabolism and oxidative phosphorylation, which are mainly responsible for releasing energy (ATP), were significantly downregulated in the females in ‘colonization’ mode compared with the females in ‘flight’ mode (Fig. 4C,H). The expression levels of seven of the differentially expressed genes involved in the TCA cycle were further validated by qPCR (Fig. S9B).

A total of 1282 genes were differentially expressed in the flight muscles between methoprene-treated and control females at 24 h after the treatment (Fig. 4E). KEGG aggregation analysis revealed the significantly upregulated and downregulated pathways (Table S5), which again included a number of genes with functions in the proteasome, in protein processing and export, and in the autophagy pathway that were significantly upregulated, whereas other genes involved in 2-oxocarboxylic acid metabolism in the TCA cycle were significantly downregulated in the flight muscles of methoprene-treated females compared with those in the acetone-treated females (Fig. 4F,H; Fig. S9C). In addition, heatCluster analysis of the 6442 differentially expressed genes showed that JH application and feeding resulted in almost consistent transcriptional changes in the flight muscles (Fig. 4G). RNA-seq experiments indicated that JH signaling inhibited the energy supply in the flight muscles of winged aphids. We then determined that the ATP levels in the flight muscles remained almost stable from 24 to 72 h post eclosion in starved aphids, whereas they were decreased upon feeding and JH application (Fig. 4I,J), which could contribute to flight-muscle degeneration. In conclusion, nutrition-dependent JH sensitivity promotes flight-muscle degeneration by suppressing the ATP supply and activating the expression of genes involved in the proteasome pathway in the flight muscles during the dispersal-reproduction transition in aphids.

First, this study emphasizes hormone sensitivity as a key regulatory mechanism of developmental events in insects. Hormone sensitivity usually refers to the expression levels of the hormone receptor and its associated proteins, as well as to the ability to respond to a hormone in a target tissue (Tang et al., 2011). During the colonization of winged aphids, nutrition-dependent gene expression levels of the JH nuclear receptor Met, an indicator of JH sensitivity, were quickly upregulated in the flight muscles; this upregulation initiates flight-muscle degeneration. Here, we demonstrate that TORC1-mediated nutritional signaling preferentially upregulates the expression of Met, leading to a tissue-specific increase of JH signaling (indicated by Kr-h1 expression) in the flight muscles (Figs 1–3). The variation in hormone sensitivity between tissues could be a general regulatory mechanism underlying tissue-specific responses to hormones. Shingleton et al. (2005) also observed organ-specific sensitivity to insulin receptor signaling during development in Drosophila. Another example is the extreme growth of beetle horns, which are highly sensitive to changes in nutrition via insulin signaling (Emlen et al., 2012) and to sex differentiation via JH signaling (Gotoh et al., 2014). It is known that TORC1-mediated nutritional signaling plays roles in the stimulation of JH biosynthesis (Maestro et al., 2009; Pérez-Hedo et al., 2013; Lu et al., 2016; Zhu et al., 2020). There was no significant upregulation in any of the JH-biosynthesis genes at the beginning of flight-muscle degeneration. The JHAMT gene has been replicated in the pea aphid genome, but only JHAMT2 was found to be regulated by the TORC1 pathway during larviposition mode (Fig. 3). In addition, the gene expression level of JHAMT2 was the highest among the JHAMT genes, and JHAMT2 was expressed in the heads in a tissue-specific manner (Fig. S6A). Thus, we deduce that the JHAMT2 homolog is the functional JHAMT gene in the pea aphid. Nevertheless, in response to nutrition availability during the dispersal-reproduction transition of winged pea aphids, JH sensitivity in the flight muscles certainly plays a more important role than JH biosynthesis for flight-muscle degeneration.

Second, this study proves that JH promotes the dispersal-reproduction life-history transition of female winged aphids (Fig. 2). In addition to the experimental data presented, it has been shown that blocking JH biosynthesis with precocene II activates flight behavior (Kobayashi and Ishikawa, 1994); most of the precocene II-treated aphids flew away from the host plants and did not show any feeding behavior; thus, they did not undergo flight-muscle degeneration (Fig. S6B,C). Nevertheless, stable JH biosynthesis is necessary for colonization behavior and the subsequent dispersal-reproduction transition of winged aphids. JH has been the main subject of endocrine regulation of dispersal-reproduction transitions in diverse insect species (Zhang et al., 2019). In response to environmental cues, such as nutrition availability, insects possess the ability to maintain metabolic homeostasis (Koyama et al., 2013; Johnson et al., 2014; Li et al., 2019b). Increasing evidence indicates that JH plays essential roles in the regulatory network of energy metabolism, with diverse and even opposing roles, depending on the developmental stage or physiological process in different insect species (Tian et al., 2010; Xu et al., 2013; Hou et al., 2015; Wang et al., 2017). Both dispersal and reproduction are costly, and loss of flight in stable habitats with available nutrition allows parthenogenetic winged aphids to shift resources toward the production of offspring away from flight. The combined data in this study suggest that a JH-induced tissue-specific reduction in energy supply mediates this transition in aphids (Fig. 4). It is worth mentioning that the mechanisms underlying the loss of flight are very different in male pea aphids, in which the wing dimorphisms are under genetic control (Li et al., 2020), and parthenogenetic females. The cricket Gryllus firmus should be an ideal model to verify whether the molecular mechanism that we discovered in the pea aphid is conserved in the regulation of flight-muscle degeneration in other insect species (Vellichirammal et al., 2014) and, moreover, to overcome the limitation of the low-efficiency of RNAi in the pea aphid.

Third, this study reveals that mitochondria might function as an important target organelle of JH signaling in the target tissue cells. Mitochondria are the primary energy-generating system in most eukaryotic cells. Mitochondrial dysfunction has pleiotropic effects on disease, aging and development (Chan, 2006). In almost all TEM analyses, we observed malformed mitochondria and mitochondrial matrix condensation, which precedes muscle-fiber breakdown (Figs 1-3). Notably, mitochondrial pathology is also a major causative factor for the earliest manifestation of muscle degeneration, as well as for muscle degeneration in Parkinson's disease (Greene et al., 2003). Impairment of the α-ketoglutarate dehydrogenase complex of the TCA cycle and respiratory chain complex I of the respiratory chain induces a decrease in ATP production and oxidative stress (Berndt et al., 2013), which could decrease the mitochondrial DNA content and trigger mitochondrial disruption, thus triggering apoptosis (Jaiswal et al., 2015; Tait and Green, 2010). Transcriptome analysis showed that JH suppresses gene expression in the TCA cycle pathway (Fig. 4), which might result in mitochondrial disruption and subsequent muscle-fiber breakdown. We have only partially verified this hypothesis using inhibitors of the TCA cycle and oxidative phosphorylation; these experiments are limited by the strong toxicity of systemic application of these chemicals.

In summary, TORC1-mediated nutritional signaling promotes JH sensitivity and thus JH signaling specifically in the flight muscles of winged aphids. The increased JH signaling inhibits the TCA cycle to reduce the energy supply and activates the expression of genes involved in the proteasome pathway, resulting in mitochondrial disruption and muscle-fiber breakdown in the flight muscles (Fig. 5). This study highlights that tissue-specific hormone sensitivity is an important regulatory mechanism of developmental transitions in insect life history.

Wingless morphs of the pea aphid, Acyrthosiphon pisum, derived from a long-established parthenogenetic clone, were kept at low density. The aphids were maintained on broad bean seedlings in a climate chamber at 18°C and ∼70% relative humidity with a 16 h:8 h (light:dark) photoperiod. Winged morphs were induced under high-density conditions as described by Guo et al. (2016). The offspring produced during a 24 h period were retained and further screened on the basis of visible wing primordia at the third instar stage. Newly emerged adults within 12 h were collected to harvest winged aphids with synchronized development as described by Kobayashi and Ishikawa (1994). In the case of starvation, winged aphids were cultured in a Petri dish containing filter paper. In the case of feeding, newly emerged or treated winged aphids were cultured within our experimental cages for colonization (Fig. S10A).

To search for pea aphid orthologues of JH-related genes involved in biosynthesis and signaling, we performed BLAST analysis (http://blast.ncbi.nlm.nih.gov/) using the protein sequences of JH-related genes of Drosophila melanogaster with the A. pisum genome. The sequences with significantly lower E-values or those predicted by automated computational analysis as presupposed genes were chosen as the putative pea aphid orthologues (HMGS1, hydroxymethylglutaryl-CoA synthase 1, gene ID 100165154; HMGS2, hydroxymethylglutaryl-CoA synthase 2, gene ID 100161670; HMGR, 3-hydroxy-3-methylglutaryl-coenzyme A reductase, gene ID 100165462; MVD, diphosphomevalonate decarboxylase, gene ID 100574505; IPPI, isopentenyl diphosphate isomerase, gene ID 100166744; FALD, farnesol dehydrogenase, gene ID 100160714; JHAMT1, juvenile hormone acid O-methyltransferase, gene ID 100569254; JHAMT2, juvenile hormone acid O-methyltransferase-like, gene ID 100160278; JHAMT3, juvenile hormone acid methyltransferase, gene ID 100166857; JHAMT4, juvenile hormone acid O-methyltransferase-like, gene ID 100572117; CYP15A1, methyl farnesoate epoxidase-like, gene ID 100162751; Met, circadian locomoter output cycles protein kaput, gene ID 100573780; Kr-h1, zinc finger X-linked protein ZXDB, gene ID 100159203). RNA-seq analysis of the heads was employed to further estimate the functional JHAMT homolog in the pea aphid. The candidate genes Met and Kr-h1 were further verified by RT (reverse transcription)-PCR and re-sequencing. Briefly, the first-strand cDNA was synthesized using the PrimeScript II 1st Strand cDNA Synthesis Kit (Takara, Dalian, China). Target-specific primers (Table S1) were synthesized to amplify the complete coding sequences (CDSs) using the first-strand cDNA as templates, and high-fidelity PrimeSTAR GXL DNA Polymerase (Takara, Dalian, China) was used in our amplification experiments according to the manufacturer's protocol. PCR-amplified fragments were directly purified and supplied for sequencing. The ORF Finder (http://www.bioinformatics.org/sms2/orffind.html) was used to predict open reading frames, and the online tools ExPASy (https://web.expasy.org/translate/) and SMART (http://smart.embl-heidelberg.de) were used to translate and identify conserved functional domains based on our resequencing results, respectively (Fig. S10B,C). The sequenced complete CDSs of Met-A (OM371056), Met-B (OM371057), Kr-h1-A (OM371058) and Kr-h1-B (OM371059) were uploaded to the National Center for Biotechnology Information (NCBI) database. In addition, we performed RNAi against Kr-h1 at the nymph stage of the pea aphid to further validate Kr-h1 orthology. RNAi against Kr-h1 at the third nymph stage resulted in abnormal wings post eclosion but no pre-metamorphosis phenotype (Fig. S10D).

The JH mimic methoprene (Sigma-Aldrich), the corpus allatum inhibitor precocene II (Yuanye Bio-Technology, Shanghai, China) and the TOR inhibitor rapamycin (Selleck, Shanghai, China) were used in this study. Methoprene (1 mg/ml) and precocene II (50 mg/ml) solutions were prepared with acetone, and 100 nl of the solution was topically applied to the aphid pronotum using a Nanoject II injector (Drummond Scientific). The aphids in flight mode were treated with methoprene, and the treated aphids were subjected to starvation or feeding for JH induction or rescue experiment, respectively; methoprene application on the femur served as a parallel control of JH induction experiment. Besides, the aphids in colonization mode were treated with methoprene and the treated aphids were subjected to feeding for another JH induction experiment. Rapamycin was first dissolved in DMSO and further diluted to 2 nM with 0.01 M PBS, and 100 nl of the solution was delivered into the abdomen via microinjection. The aphids in flight mode were treated with rapamycin or precocene II using the Nanoject II injector, and the aphids were placed in the feeding cages after the treatment. At least 30 aphids were treated for each group every time.

qPCR was employed to detect the changes in JH biosynthesis and signaling levels during A. pisum winged adult colonization or after chemical or RNAi treatment. The spatial expression profiles of Met and Kr-h1 were determined using tissues from the heads (30 individuals for one biological replicate), thoraxes (10 individuals), ovaries (10 individuals), fat body (20 individuals), flight muscles (10 individuals) and legs (50 individuals) of winged adults at feeding or starvation 60 h after eclosion. The temporal expression profiles of JHAMTs, Met and Kr-h1 at different developmental stages of winged adults subjected to feeding or starvation conditions post eclosion were determined using head samples for JHAMTs and thorax samples for Met and Kr-h1. For the topical application of methoprene, starved winged aphids were treated at 24 h post eclosion and the thoraxes were excised from the specimen at 2, 6, 12, 24 and 48 h after the treatment; whereas feeding winged aphids were treated 48 h post eclosion and the heads, thoraxes, ovaries, fat body, flight muscles and legs were excised from the specimens 12 h after the treatment for qPCR analysis. For RNAi and rapamycin application experiments, the heads were excised for JHAMT detection and the flight muscles or fat body were excised for Met and Kr-h1 detection at specific points in time. Three independent biological replicates were performed in all qPCR analyses. All of the samples were stored in RNAiso Plus (Takara) at −80°C until use. Following total RNA isolation and reverse transcription, qPCR was performed on an IQ-5 Real-time system (Bio-Rad). The expression levels of A. pisum elongation factor 1 α (EF1a) and ribosomal protein S20 (Rps20) were quantified as housekeeping genes for normalization (Chen et al., 2016; Ishikawa et al., 2012). The cycle threshold values normalized against the means of EF1a and Rps20 standard values were obtained to calculate the quantitative variation of target genes using the mathematical model by Pfaffl (2001) and the REST 2009 software (QIAGEN). Gene-specific qPCR primers (targeting the common regions of transcript variants) with appropriate amplification efficiency were screened by cDNA gradient-based standard curves (Table S2).

Four groups of flight-muscle samples and two groups of head samples (each group containing three independent biological replicates) were prepared for RNA sequencing. In the feeding condition, winged adults in ‘flight’ and ‘colonization’ modes were subjected to flight-muscle (48 h post eclosion) or head (60 h post eclosion) sampling. In the starvation condition, acetone- and methoprene-treated aphids were subjected to flight-muscle sampling at 24 h after the treatment. Each flight-muscle or head sample was collected from at least 30 aphids and was prepared for RNA extraction. An aliquot of 1000 ng of total mRNA from each sample was used to construct libraries. Single-end, 1×50 sequencing was performed on an Illumina HiSeq 2500 at the Novogene sequencing facility (Beijing, China). The genome sequence and annotation files of A. pisum were downloaded from the NCBI database (https://www.ncbi.nlm.nih.gov/genome/?term=pea+aphid). The clean reads were mapped to the genome using genomic short-read nucleotide alignment program (GSNAP) to estimate the expression level of all transcripts (Wu et al., 2005). Gene expression levels were estimated by fragments per kilobase of transcript per million fragments mapped. Differential expression analysis of two groups was performed using the DESeq R package (1.10.1). Genes with an adjusted P-value (false discovery rate, FDR)<0.05 found by DESeq were designated as differentially expressed. We used KOBAS (Mao et al., 2005) software to test the statistical enrichment of differentially expressed genes in KEGG pathways. The entire RNA-seq datasets are available online (PRJNA828046).

Gene-specific primers (Table S3) were designed to amplify a 307 bp fragment (common regions of isoforms A and B) for Kr-h1. A 193 bp fragment of the Mus musculus lymphotoxin A (Lta) was also cloned to serve as an unrelated control gene. The validated plasmid and primers containing T7 RNA polymerase promoter were subjected to PCR amplification to produce DNA templates used for double-stranded RNA (dsRNA) synthesis. DsRNA synthesis and purification were performed using a T7 RiboMAX Express RNAi System (Promega), according to the manufacturer's instructions. The concentration of the purified dsRNA was quantified at 260 nm using a NanoDrop 2000c spectrophotometer (Thermo Fisher Scientific), and was adjusted to 6 µg/µl with RNase-free water. Newly emerged aphids were subjected to 24 h of fasting prior to dsRNA microinjection. They were anesthetized with carbon dioxide, and 150 nl of specific dsRNA was injected into the aphid hemolymph using a Nanoject II injector. The dsRNA-treated aphids were placed in the feeding cages for colonization. At least 30 aphids were treated for each group every time.

Newly collected muscle tissues were fixed in 2.5% glutaraldehyde at 4°C overnight followed by thorough washing with PBS. The specimens were then postfixed in 0.5% osmium tetroxide for 2 h, dehydrated in an ethanol gradient and finally embedded in Epon resin. Longitudinal sections (70 nm) of muscle fibers were prepared using a UC7 ultramicrotome (Leica), stained with Reynold's lead citrate and visualized on an HT7700 transmission electron microscope (Hitachi, Tokyo, Japan) to observe the degree of breakage of muscle fibers and the morphology of mitochondria. Quantitative analysis of the damaged muscle fibers was performed by ImageJ.

For each sample, 30 aphids without embryos were lysed directly for total protein extraction. Proteins were separated by 12% SDS-PAGE and transferred to a polyvinylidene fluoride membrane. The loaded membrane was then blocked with 5% bovine serum albumin in TBS supplemented with 1% Tween-20 (TBST, pH 8.0) for 1 h at room temperature and incubated with 1:1000 diluted primary antibodies (anti-phospho-4EBP, Thr37/46, 2855S, Cell Signaling Technology; and anti-tubulin, AF1216; Beyotime Biotechnology, China) at 4°C overnight. The primary antibodies specifically bind to target proteins. Then, the membrane was incubated with goat anti-rabbit IgG-HRP secondary antibody (1:5000; A0181, Beyotime Biotechnology) for 1 h at room temperature. The immune signal was visualized by the ECL chemiluminescent substrate (Boster, Wuhan, China), and images were captured on a ChemiDoc MP System (Bio-Rad). Quantitative analysis of the protein bands was performed by ImageJ.

ATP levels in the flight muscles were determined using an ATP Assay Kit (Fluorometric) (ab83355, Abcam). A standard curve was established from a pure ATP gradient according to the manufacturer's instructions. In each biological replicate, approximately 10 mg of the aphid thoraxes was homogenized in 2 N perchloric acid and the diluted supernatant was neutralized with 2 M KOH. These deproteinized samples and ATP reaction mixes were successively added into a black 96-well plate, and the reaction was incubated at room temperature for 30 min. The fluorescence readings were measured by an Infinite 200 microplate reader (Tecan, Männedorf, Switzerland) with the excitation/emission setting at 535/587 nm. The mean fluorescence intensity was determined by comparison to a standard curve.

The authors are grateful to Prof. Michael R. Strand (University of Georgia), Yun-Xia Luan (South China Normal University, China) and Dr He-He Cao (Northwest A&F University, China) for critical revision of the manuscript, and to the Sagene Biotech Co., Ltd. (Guangzhou, China) for the schematics of aphids and plants used in Fig. 5.