Transcriptome Analysis of the Brain and Gnathal Ganglion Reveals Feeding-Mediated Genes in Helicoverpa armigera Larvae

TL;DR

This study identifies genes in the central nervous system of Helicoverpa armigera larvae that are linked to feeding behavior, offering new targets for pest control.

Contribution

The first systemic transcriptomic analysis of feeding-related genes in the central nervous system of Helicoverpa armigera larvae.

Findings

944 differentially expressed genes were identified in the brain and gnathal ganglion of H. armigera larvae.

41 candidate genes, including neuropeptides and gustatory receptors, were found to be closely associated with feeding behaviors.

Most candidate genes were upregulated during the fifth instar stage and under starvation conditions.

Abstract

Despite the central role of the gustatory nervous system in regulating feeding behavior in the larvae of the major agricultural pest Helicoverpa armigera, the systemic molecular basis of this process is largely unknown. To investigate the molecular basis, we performed RNA-seq on dissected brains and gnathal ganglion (GNG) from fifth-instar larvae, revealing 944 differentially expressed genes (DEGs) that are potentially involved in feeding behaviors. Bioinformatic analyses revealed significant enrichment of these DEGs in pathways including “taste transduction”, “neuroactive ligand–receptor interaction”, and “feeding behavior”. Furthermore, 41 candidate genes closely associated with feeding behaviors were screened, including neuropeptides, neuropeptide receptors, gustatory receptors, and feeding-mediated proteins. Phylogenetic analyses demonstrated that these key genes are evolutionarily conserved. Quantitative real-time PCR (qPCR) results confirmed that most of these candidate genes were significantly upregulated during the fifth instar stage, and their expression was further enhanced under starvation conditions (p < 0.05). Our study presents the first systemic feeding-mediated transcriptomic profile in the insect central nervous system and identifies key feeding-mediated genes in H. armigera larvae. These findings advance our understanding of the molecular basis of feeding in Lepidopteran pests and offer potential target genes for developing gustatory-based pest control strategies.

Figures

Full text

1. Introduction

Feeding behavior is a core life-sustaining activity in insects, which is essential for survival, growth, development, and reproduction. This process can be divided into four sequential stages: the initiation of feeding motivation, food source searching, nutrient recognition, and food ingestion [1,2]. Each stage is precisely regulated by the CNS—the brain and the gnathal ganglion (GNG)—ensuring that the insects efficiently acquire nutrients, maintain metabolic homeostasis, and support normal growth, development, and reproductive processes.

Hunger serves as a key internal state driving feeding behavior. When nutrients or energy are deficient, internal sensory systems in insects—such as insulin-producing cells, the corpora cardiaca, and the fat body—detect changes in indicators such as blood glucose levels, thereby initiating foraging behavior via neuroendocrine signaling. For example, hunger suppresses the secretion of Unpaired 2 (Upd2) from the fat body, which in turn reduces the release of insulin-like peptides (ILPs) and promotes food-seeking behavior [3,4]. Simultaneously, the secretion of adipokinetic hormone (AKH) increases, enhancing the insect’s locomotor capacity to facilitate foraging [5]. ILP and AKH signaling exhibit mutually antagonistic regulation, establishing a dynamic hormonal balance under hunger conditions [6]. Upon satiety, Upd2 activates the JAK/STAT pathway to inhibit GABAergic neurons, thereby relieving the suppression of insulin-producing cells, promoting ILP release, and further inhibiting feeding [3]. Additionally, neuropeptides such as myoinhibitory peptide (MIP) and allatostatin A (Ast-A) participate in satiety signaling, cooperatively suppressing feeding behavior [7,8,9].

The initiation and termination of feeding behavior are dynamically regulated by a series of neuropeptides, which play critical roles in modulating key physiological processes such as feeding and metabolism [9,10,11,12,13]. Based on their functions, these signals can be categorized into two main classes: hunger signals and satiety signals. Under hunger conditions, signals such as neuropeptide F (NPF), short neuropeptide F (sNPF), and AKH are rapidly released, promoting feeding and enhancing foraging activity. In contrast, after satiety, signals including ILPs, Ast-A, and MIP are upregulated, effectively suppressing feeding [3,4,5,7,8].

During the food-seeking stage, the olfactory system exhibits significant plasticity in response to hunger. Hungry insects show increased sensitivity to food odors and reduced avoidance responses to toxic odors. In Drosophila melanogaster, hunger lowers ILP signaling while upregulating the expression of sNPF and tachykinin receptors. This enhances the response of OR42b neurons to attractive odors and suppresses the response of OR85a neurons to aversive odors, thereby optimizing foraging efficiency [14]. A similar mechanism has been observed in Bactrocera dorsalis, where sNPF signaling enhances the sensitivity of olfactory neurons to food odors [15]. Conversely, satiety signals such as MIP promote avoidance behavior toward food odors [8].

Although the olfactory system guides food localization, the final feeding decision relies on the taste system’s evaluation of the nutritional value and safety of food. Hunger dynamically modulates taste sensitivity: sweet perception is enhanced, while bitter perception is reduced. This process is regulated by the neuropeptides NPF and sNPF. Under hunger conditions, NPF enhances the response of Gr5a sweet-sensing neurons via dopaminergic signaling, while sNPF suppresses the sensitivity of Gr66a bitter-sensing neurons, both acting synergistically to promote feeding [16,17]. AKH may also participate in the regulation of bitter sensitivity through the sNPF pathway, though its direct mechanism remains unclear [17].

The food ingestion stage involves the coordination of mouthpart movement and continuous pharyngeal sensing. Pharyngeal gustatory neurons expressing Gr43a and Gr64e detect ingested sugars, thereby supporting sustained feeding [18], whereas neurons expressing Gr33a inhibit feeding by detecting substances such as caffeine [19]. The GNG serves as a key sensorimotor integration center, in which specific neurons can directly respond to sucrose signals to drive feeding behavior [20]. Moreover, interoceptive subesophageal zone neurons (ISN), which express AKH receptors (AKHR) and the Nanchung, integrate metabolic and water-balance signals to adjust the food-to-water intake ratio, enabling coordinated regulation of nutritional and water homeostasis during the ingestion stage [21].

The CNS of insects functions as the principal regulatory hub governing feeding behavior, coordinating key processes such as feeding motivation, foraging, and ingestion through a variety of neuroactive substances, including hormones, neuropeptides, and biogenic amines [22,23]. The cotton bollworm, Helicoverpa armigera, is a globally significant agricultural pest that causes substantial economic losses annually [24,25]. Its larvae exhibit strong feeding capacity and broad host adaptability, indicating the presence of a highly sophisticated gustatory system. This system is primarily composed of the brain and the distinct, separate GNG, where the latter receives and processes gustatory information via peripheral nerve bundles before transmitting it through the circumesophageal connective (CC) to the brain for higher-order integration and decision-making [26,27]. Although the functions of peripheral taste organs in detecting specific compounds have been confirmed [28,29,30], and several peripheral gustatory receptors have been identified [31,32,33], systematic studies on feeding-mediated genes in the brain and GNG of H. armigera larvae remain limited, and the mechanisms underlying signal integration and behavioral coordination are still unclear. To address this knowledge gap, we adopted an integrated approach combining transcriptomics, bioinformatics, and phylogenetic analyses to systematically screen and identify feeding-mediated genes in the brain and the GNG of H. armigera larvae.

2. Materials and Methods

2.1. Insect Rearing

The H. armigera used in this study originated from a wild population collected at the Xinxiang Comprehensive Experimental Base of the Chinese Academy of Agricultural Sciences. A stable laboratory colony was established and reared on an artificial diet under controlled conditions: 26 ± 1 °C, 70 ± 5% rh, with a 14:10 h (l:d) photoperiod. Fifth-instar larvae at day 2 were used as the experimental subjects, while adults were maintained for colony expansion. Upon eclosion, adults were provided daily with a 10% sucrose solution supplemented with a vitamin C complex. The larvae’s artificial diet was prepared according to the following formulation: wheat bran (150 g, Boxiang Biotechnology Co., Ltd. Zhengzhou, China), soybean meal (80 g, Boxiang Biotechnology Co., Ltd. Zhengzhou, China), yeast powder (25 g, Boxiang Biotechnology Co., Ltd. Zhengzhou, China), casein (40 g, SBoxiang Biotechnology Co., Ltd. Zhengzhou, China), sorbic acid (3 g, Sinopharm Chemical Reagent Co. Ltd., Shanghai, China), ascorbic acid (3 g, Sinopharm Chemical Reagent Co. Ltd., Shanghai, China), methylparaben (6 g, Sinopharm Chemical Reagent Co. Ltd., Shanghai, China), sucrose (10 g, Sinopharm Chemical Reagent Co. Ltd., Shanghai, China), agar (20 g, Sinopharm Chemical Reagent Co. Ltd., Shanghai, China), and a vitamin premix (0.8 g, Sinopharm Chemical Reagent Co. Ltd., Shanghai, China). These ingredients were thoroughly mixed and dissolved in 1600 mL of distilled water to prepare the final diet.

2.2. Transcriptome Sequencing

2.2.1. Tissue Dissection, RNA Extraction and Quality Control

Sample Preparation: Two-day-old fifth-instar larvae with uniform size were selected for dissection. Both brain tissue and the GNG were rapidly dissected in ice-cold sterile PBS buffer (Figure 1). Dissected tissues from each group (n ≥ 50) were pooled, immediately snap-frozen in liquid nitrogen, and stored at −80 °C until RNA extraction. Total RNA was extracted using Trizol reagent according to the manufacturer’s instructions. RNA integrity and concentration were assessed using an Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA).

2.2.2. Library Preparation

Libraries were constructed using the NEBNext Ultra II RNA Library Prep Kit following the manufacturer’s instructions. Briefly, approximately 1 μg of total RNA per sample was used. Poly(A)-tailed mRNA was enriched using Oligo dT magnetic beads. The qualified libraries were quantified using a Qubit 4.0 Fluorometer (Thermo Fisher, Waltham, MA, USA), and their size distribution was verified using the Agilent 2100 Bioanalyzer. The effective library concentration was accurately determined by qPCR using the KAPA Library Quantification Kit (Roche, Basel, Switzerland). Equimolar amounts of each library were pooled and subjected to paired-end sequencing (2 × 150 bp) on an Illumina NovaSeq 6000 platform.

2.2.3. Data Quality Control

A stringent bioinformatic pipeline was employed for quality control of the raw sequencing data. Initially, raw reads were processed using FastP (v0.23.1) to efficiently identify and remove low-quality sequences. Specific filtering steps included (1) removal of reads containing adapter sequences; (2) exclusion of reads with poly-N content exceeding 10%; and (3) filtration of low-quality reads (where over 50% of bases had a Phred quality score < 20). The resulting high-quality clean reads were used for subsequent analyses. To comprehensively assess data quality, multiple metrics were calculated for the clean data, including the percentages of bases with Q20 (base call accuracy ≥ 99.0%) and Q30 (base call accuracy ≥ 99.9%), as well as GC content and its distribution.

2.2.4. Read Alignment to the Reference Genome

The reference genome and corresponding gene model annotation files were directly downloaded from the NCBI (National Center for Biotechnology Information) genome database at https://www.ncbi.nlm.nih.gov/datasets/genome/GCF_023701775.1/ (accessed on 10 June 2022). GenBank assembly is GCA_023701775.1. The annotation is NCBI Helicoverpa armigera Annotation Release 101. Clean reads were then aligned to the reference genome using HISAT2 v2.0.5, and gene expression levels were quantified. Novel gene prediction was performed with StringTie (v1.3.3b), while read counts mapped to each gene were quantified using featureCounts (v1.5.0-p3). Differential expression analysis between the two comparative groups was conducted with DESeq2 (v1.20.0).

2.2.5. Functional Enrichment Analysis

GO and KEGG enrichment analyses of the differentially expressed genes were subsequently carried out using the clusterProfiler (v3.8.1). Differentially expressed genes (DEGs) were identified based on set thresholds. All DEG results were adjusted for multiple testing using the False Discovery Rate (FDR) method (q-value ≤ 0.05) and |Log_2_ (Fold Change)| ≥ 1.0. GO enrichment analysis was performed to identify significantly overrepresented terms in the DEGs set compared to the background gene set across the biological process, molecular function, and cellular component categories. Simultaneously, KEGG pathway enrichment analysis was conducted to reveal significantly enriched metabolic and signal transduction pathways. All enrichment results were adjusted for multiple testing using the FDR method (q-value ≤ 0.05) and visualized to illustrate key enriched functional categories and pathways.

2.3. Identification of Candidate Feeding-Mediated Genes

GO and KEGG enrichment analyses were initially performed to identify key biological processes and pathways. Then, based on the resulting foundation and the existing literature, a set of candidate genes highly associated with the mediated mechanisms of feeding behavior was screened. The identified candidate genes were then compared with those of representative Lepidopteran insects, such as Bombyx mori and Manduca sexta, as well as established model insects, including Locusta migratoria and D. melanogaster, to determine their homologous relationships. Next, the expression levels of target genes across samples were evaluated based on FPKM values. Finally, by incorporating known functional annotation information, the target genes were confirmed and selected.

2.4. Construction of the Phylogenetic Tree

Amino acid sequences of feeding-mediated genes from H. armigera and other related insect species were obtained from NCBI. Multiple sequence alignment was performed with Clustal Omega (v1.2.2). A maximum-likelihood tree was reconstructed using FastTree v2.1.11 with 1000 ultrafast bootstrap replicates. Branches with bootstrap support ≥70% were considered robust. The final tree was visualized and annotated in iTOL.

2.5. Quantitative Real-Time PCR Analysis

Total RNA was extracted from multiple larval tissues of H. armigera, including the brain (Br), gnathal ganglion (GNG), thoracic ganglion (TG), abdominal ganglion (AG), fat body (Fb), and midgut (Mg). RNA was also isolated from larval heads at different developmental stages: the second day of each instar (L1-2, L2-2, L3-2, L4-2), and days 0, 2, and 5 of the fifth instar (L5-0, L5-2, L5-5), plus starved fifth-instar day-2 larvae (L5-2-ST; the starvation period was 1 day). The method of RNA extraction was the same as that in Section 2.2.1. For each sample, 1 μg of total RNA was treated with gDNA Eraser to remove genomic DNA contamination, followed by reverse transcription according to the manufacturer’s instructions, yielding high-quality cDNA templates.

To quantify target gene expression, qPCR was conducted using Ha-Actin as a stable housekeeping reference. Primer sequences for the 41 candidate feeding-mediated genes are listed in Table S1. Expression levels were calculated via the 2^−ΔΔCT^ method, with three biological replicates (each with three technical replicates) per sample group.

2.6. Data Analysis

Statistical analysis was performed using SPSS 21.0. For the gene expression differences between normal and starvation, the t-test was used as the statistical analysis method. The KEGG and heatmap figures were generated with Excel (2021) and R 4.3.2. All bar charts were generated using GraphPad Prism 9.1 software. All tables were generated with Excel (2021).

3. Results

3.1. Quality Control of Transcriptome Data and Inter-Sample Correlation Analysis

Transcriptome analysis was performed on the brain and GNG dissected from two-day-old fifth-instar larvae (Figure 1A–C). Six cDNA libraries were constructed and sequenced on the Illumina NovaSeq 6000 platform. Raw reads were processed to remove low-quality sequences, resulting in 38.84 GB of clean data. These data showed high sequencing quality, with Q20 and Q30 values reaching 97% and 91%, respectively, and GC content ranging from 42.51% to 45.52% (Table S2).

The Pearson correlation coefficient was used to evaluate intra- and inter-group relationships (Figure 1D). All squared correlation coefficients (R^2^) exceeded 0.8, demonstrating strong biological reproducibility. Weighted gene co-expression network analysis (WGCNA) classified the genes into three distinct modules (blue, turquoise, and brown), as shown in the module eigengene adjacency heatmap. Subsequent analysis identified key hub genes within each module (Table S3).

3.2. Analysis of Differentially Expressed Genes in the Brain and Gnathal Ganglion

We adopted a two-step strategy to identify differentially expressed genes (DEGs) between GNG and brain tissues. Initial screening (P_adj_ ≤ 0.05, |Log_2_ FC| ≥ 1.0) identified 944 DEGs (465 up- and 479 down-regulated), whose expression distribution is shown in Figure S1. Hierarchical clustering of these DEGs revealed tissue-specific profiles (Figure 2A). Applying a stricter threshold (|log_2_ FC| ≥ 2.0) refined the set to 297 high-confidence DEGs (120 strongly upregulated, 177 strongly downregulated) (Figure 2B,C; Table S4).

3.3. GO and KEGG Enrichment Analyses of Differentially Expressed Genes

To elucidate the biological functions of DEGs between the brain and GNG, GO and KEGG enrichment analyses were performed. DEGs highly expressed in the brain were first identified and functionally annotated. GO analysis revealed significant enrichment in several categories: biological processes including transmembrane transport, transcription regulation, and carbohydrate metabolism; cellular components such as transporter complexes and extracellular regions; and molecular functions like transmembrane transporter activity, calcium ion binding, and DNA-binding transcription factor activity (Figure 2D). Complementary KEGG analysis further indicated enrichment in key pathways, including oxidative phosphorylation, lysosomal activity, insect hormone biosynthesis, and Hippo signaling (Figure 2E). Subsequently, DEGs highly expressed in the GNG were analyzed. GO enrichment highlighted biological processes such as signal transduction and developmental process; cellular components including extracellular region and myosin complex; and molecular functions like chitin binding, ion channel activity, and G-protein-coupled receptor activity (Figure 2F). KEGG analysis showed significant association with key pathways, including Wnt signaling, FoxO signaling, carbon metabolism, and neuroactive ligand–receptor interaction (Figure 2G).

An integrated functional enrichment analysis was performed on all DEGs. Among the 738 GO terms examined, 38 terms comprising 206 genes were significantly enriched (P_adj_ ≤ 0.05). Within the biological processes category, 22 terms were significantly enriched, including “ion transport” and “regulation of transcription, DNA-templated”, indicating potential roles in ion homeostasis and transcriptional regulation. In the cellular component category, significantly enriched terms included “extracellular region”, “plasma membrane”, “plasma membrane part”, and “plasma membrane protein complex”, suggesting involvement in extracellular matrix dynamics and membrane-associated structures. For molecular function, 12 terms were significantly enriched, including “DNA binding”, “transmembrane transporter activity”, and “transporter activity”, indicating roles in DNA-related mediated processes and transmembrane transport (Figure 3A). KEGG pathway enrichment analysis identified 101 significantly enriched metabolic pathways (P_adj_ ≤ 0.05), among which “Oxidative phosphorylation” showed the highest significance (Figure 3B). These DEGs were predominantly enriched in three key functional categories: (1) energy metabolism-related pathways (e.g., oxidative phosphorylation and glycolysis/gluconeogenesis); (2) signal transduction pathways (e.g., Wnt signaling pathway, mTOR signaling pathway, neuroactive ligand–receptor interaction); and (3) fundamental metabolic pathways (e.g., pyrimidine metabolism, glyoxylate and dicarboxylate metabolism, and glycine/serine/threonine metabolism). The enrichment pattern suggests that the DEGs may regulate physiological functions like feeding behavior and development through energy metabolism, neural signal transduction, and nutrient metabolism. Additionally, WGCNA was performed on all DEGs. Sample clustering revealed no obvious outliers, confirming data integrity (Figure 3C). To construct a scale-free co-expression network, we evaluated soft-thresholding powers. At a power of 2, the scale-free topology fit index (R^2^) exceeded 0.85, meeting the empirical criterion for a scale-free network, while mean connectivity decreased as expected (Figure 3D). To balance scale-free topology with biological interpretability, a power value of 2 was selected for network construction, yielding two distinct co-expression modules as shown in the hierarchical clustering dendrogram (Figure 3E).

3.4. Identification and Analysis of Feeding-Mediated Genes in the Brain and Gnathal Ganglion

From our dataset, we examined 41 candidate genes potentially involved in feeding regulation. Among them, 20 were DEGs, while the remaining 21 showed no significant change in expression. These 20 feeding-mediated DEGs encoded products, including neuropeptides (SIFamide, Corazonin) and their receptors (SIFamide receptor, Neuropeptide F receptor), as well as feeding-associated proteins such as Trehalose transporter Tret1 and Homeobox protein ceh-19 (Table 1). Further screening of co-expressed genes identified additional candidates encoding neuropeptide receptors (e.g., CCHamide-1 receptor, Neuropeptide F receptor), neuropeptides (e.g., Allatostatin, Short neuropeptide F), and gustatory receptors (Gustatory receptors 28a, 68a, and 5a). Notably, several gustatory receptors exhibited extremely low or barely detectable expression levels (Table 1). Functional enrichment analyses provided further insight. For the differentially expressed feeding-mediated genes, GO analysis indicated significant enrichment in biological processes such as hormone activity, receptor ligand activity, and taste sensory perception (Figure 4A). Corresponding KEGG analysis revealed strong associations with longevity regulation, mTOR signaling, and FoxO signaling pathways (Figure 4B). In contrast, the non-differentially expressed feeding-mediated genes displayed distinct patterns. GO analysis highlighted transmembrane signaling receptor activity and sensory perception functions (Figure 4C), while KEGG enrichment pointed to neuroactive ligand–receptor interactions and folate biosynthesis (Figure 4D). These patterns suggest that although both groups are involved in feeding regulation, the DEGs may be more directly involved in modulating hormonal signaling and metabolic pathways, whereas the non-differentially expressed genes may play more constitutive roles in basic sensory perception and receptor-mediated signaling.

3.5. Construction of the Phylogenetic Tree and Expression Profiling for Feeding-Mediated Genes

In this study, four types of feeding-mediated genes were identified, including neuropeptides, neuropeptide receptors, gustatory receptors, and feeding-mediated proteins. To investigate the evolutionary history of feeding-mediated genes, we performed maximum-likelihood phylogenetic analysis using predicted amino acid sequences, with homologous sequences from related Lepidopterans (provided in Table S5). To characterize their spatiotemporal expression, we performed qPCR across key developmental stages (L1-2 to L5-5, including starved L5-2-ST) and tissues (Br, GNG, TG, AG, Fb, Mg) of fifth-instar day-2 larvae and then systematically analyzed these expression profiles. Primer sequences are listed in Table S1.

3.5.1. Neuropeptides

Based on our transcriptomic data, we identified ten neuropeptide genes. Phylogenetic analysis indicates that these neuropeptides are evolutionarily conserved, each likely playing a distinct role in feeding regulation (Figure 5A). These genes exhibited tissue-specific expression, with transcript levels significantly higher in the brain and GNG than in peripheral tissues like the fat body and midgut, where expression was low or undetectable (Figure 5B). Across larval development, expression peaked during active feeding stages but declined in the fourth instar and pre-pupal stage (L5-5). Notably, starvation strongly upregulated their expression (Figure 5C and Figure S2), suggesting that these neuropeptides may function as critical hunger signals to promote food-seeking motivation and compensatory feeding upon nutrient deprivation.

3.5.2. Neuropeptide Receptors

Neuropeptide receptors play pivotal roles in mediating neuropeptide signaling, which is integral to the regulation of feeding behaviors in insects. In this study, we systematically identified neuropeptide receptor genes from the transcriptomic data. Phylogenetic analysis indicated evolutionary conservation, suggesting their conserved roles in insect physiology, including energy metabolism, stress responses, and feeding behavior regulation (Figure 6A). Expression profiling revealed distinct tissue specificity, with predominant expression in the brain and GNG and minimal to undetectable levels in peripheral metabolic tissues like the fat body and midgut (Figure 6B). Developmentally, transcript levels were elevated in the actively feeding fifth instar and were significantly upregulated by starvation (Figure 6C and Figure S2).

3.5.3. Gustatory Receptors

We identified nine putative gustatory receptor (Gr) genes. To infer their functional classification and evolutionary relationships, we constructed a maximum-likelihood phylogenetic tree incorporating Gr sequences from representative insect species (B. mori, M. sexta, Helicoverpa zea, Spodoptera frugiperda, Pieris napi, etc.) (Figure 7A). Based on their predicted ligand specificity, the identified Grs are categorized into three groups: sweet receptors (Gr5a, Gr64a, and Gr43a), likely involved in sugar perception; bitter receptors (Gr28a and Gr24), potentially mediating aversive responses to toxic compounds; and a pheromone-sensitive receptor (Gr68a), which may contribute significantly to social or reproductive chemosensation. These genes exhibited distinct expression profiles: Gr28a was most abundant in the fat body, whereas other Grs were primarily expressed in neural tissues (brain, GNG) with minimal midgut expression (Figure 7B). The highest expression was observed during the feeding fifth instar, followed by a decrease in the pre-pupal stage. This coordinated transcriptional response may be modulated by systemic hormonal signals reflecting energy deficit, thereby enhancing gustatory sensitivity and motivating food-seeking behavior (Figure 7C and Figure S3).

3.5.4. Feeding-Mediated Proteins

Feeding-mediated proteins refer to a class of signaling molecules and functional mediators involved in the cascade of feeding behavior. These proteins integrate internal energy-state signals with external environmental cues to coordinately modulate feeding behavior in insects. A total of 15 feeding-mediated protein genes were identified, like Trehalose transporter 1 (Tret1), Glutamate decarboxylase (Gad 1), Neuronal acetylcholine receptor (NaR), 6-pyruvoyl tetrahydrobiopterin synthase (6-PTP), and others. Tret1 is primarily involved in trehalose transport, playing a critical role in maintaining hemolymph glucose homeostasis in insects. Gad1 modulates insect feeding motivation and promotes satiety by activating specific GABAergic circuits in the brain and GNG. Their evolutionary relationships were inferred via maximum-likelihood phylogeny with orthologous genes from other species, including M. sexta, Plutella xylostella, B. mori, and Ostrinia furnacalis (Figure 8A). These genes exhibited neural-enriched expression (brain, GNG, thoracic/abdominal ganglion) versus low expression in the fat body and midgut (Figure 8B). Developmentally, some showed peak expression in the first instar, declined to a low in the fourth instar, and were strongly induced by starvation in the fifth instar (Figure 8C and Figure S3).

4. Discussion

Feeding behavior is crucial for the survival, growth, development, and reproduction of insects. It enables insects to effectively identify and ingest suitable nutrients and maintain internal metabolic balance, thereby supporting their growth, development, and reproductive success and sustaining the population [34].

In the present study, we identified a comprehensive set of feeding-mediated genes in the brain and GNG of H. armigera larvae via transcriptomics. Integrated analysis showed these genes participate in gustatory detection, neuropeptide signaling, and energy metabolism. Functional divergence was evident: brain DEGs were enriched for integrative functions, while GNG DEGs were linked to sensory-motor roles (Figure 2D–G). Feeding-mediated genes were further associated with hormonal/chemosensory crosstalk (Figure 3A) and key pathways like mTOR and FoxO signaling (Figure 3B), revealing their connections to conserved networks governing energy and physiological homeostasis.

At the nutrient-sensing level, insects can monitor internal nutritional status in real time through peripheral and central chemosensory mechanisms. For instance, in D. melanogaster, the brain-expressed Gr43a receptor senses fructose levels in the hemolymph and bidirectionally regulates sucrose feeding behavior based on hunger or satiety states, indicating that central nutrient-sensing systems directly modulate feeding decisions [35,36,37]. A similar mechanism has been observed in H. armigera, where the Gr9 is highly expressed in the foregut and abdomen and participates in sensing internal fructose to promote feeding, suggesting that this internal sugar-sensing pathway may be conserved across insect species [38]. In addition to sugars, amino acids constitute another key category of nutrients. Their perception involves the coordinated action of amino acid transporters and the mTOR/S6K signaling pathway, which collectively regulate amino acid uptake and utilization, thereby maintaining protein metabolic balance [39,40,41].

The initiation and termination of feeding behavior are dynamically regulated by a series of neuropeptides, including NPF, sNPF, AKH, ILPs, Ast A, MIP, SIFa, CCHa1, CCHa2, and CRZ. These neuropeptide regulatory pathways exhibit broad conservation across insects. For example, ILPs and NPF/sNPF have been shown to regulate feeding in multiple insect species such as Maruca vitrata [42], Acyrthosiphon pisum [13,43], Apis mellifera [44], Dendroctonus armandi [12], Bactrocera dorsalis [15], and L. migratoria [45,46]. Meanwhile, other neuropeptides such as SIFa and CCHa2 play critical roles in species like Nilaparvata lugens [47], M. sexta [48], A. pisum [49], Gryllus bimaculatus [50], and Rhodnius prolixus [11], further highlighting the diversity and evolutionary complexity of neural regulatory networks underlying insect feeding behavior.

In this study, we identified a set of candidate genes encoding neuropeptides (including NPF, ILP3, CRZ, and SIFa) and their receptors, which are conserved across insect species (Figure 5 and Figure 6). The number of neuropeptides and their receptors identified in the CNS of H. armigera larvae was lower than that reported previously in species such as Tribolium castaneum [51], B. mori [52], D. melanogaster [53], N. lugens [54], A. mellifera [55], and Athyma hetaohei [56]. This discrepancy is likely primarily due to limitations associated with CNS-specific transcriptomic data.

The peripheral gustatory system plays a key role in food recognition and quality assessment in insects. For example, gustatory receptors exhibit clear functional specialization and cooperation: for example, sugar receptors such as Gr5a and Gr64f mediate feeding responses, while bitter receptors such as Gr66a and Gr33a mediate feeding aversion [57,58]. Interestingly, some receptors, including Gr5a, Gr61a, and Gr64f, can sense low concentrations of amino acids to promote feeding, whereas Gr66a participates in the inhibitory response triggered by high concentrations of amino acids [59]. Receptor co-expression (e.g., Gr43a and Gr64a) may alter neuronal response properties, suggesting the presence of complex signal integration mechanisms [60]. Importantly, hunger and satiety can bidirectionally modulate the sensitivity of gustatory neurons via neuropeptides such as NPF and sNPF: under hunger conditions, NPF enhances sweet responses through dopaminergic signaling, while sNPF suppresses bitter responses, synergistically promoting feeding. In contrast, these pathways are inhibited upon satiety [16,17]. This regulatory pattern is also observed in D. melanogaster larvae: during early foraging, Gr43a and Gr66a regulate responses to sugars and bitter compounds, respectively; after satiety, CCHa1 secreted by the fat body acts coordinately with elevated insulin levels in the hemolymph to induce feeding cessation and departure from the food source, thereby completing the transition from foraging to satiety.

Additionally, in this study, our expression profiling revealed high expression levels of key regulatory genes, including neuropeptides, their receptors, gustatory receptors, and feeding-related proteins, in the brain and GNG (Figure 5, Figure 6, Figure 7 and Figure 8). Taking neuropeptides and their receptors as an example, these molecules are predominantly expressed in the brain and GNG, indicating specialized roles in coordinating neural and endocrine signaling during larval development. Notably, under short-term starvation treatment, their expression levels showed a significant upward trend, suggesting that their regulation of feeding may be based on an acute response to nutritional status. The expression profiles of these neuropeptides closely corresponded with observed feeding behaviors, thereby underscoring their unique and critical roles in both promoting and inhibiting feeding. These coordinated expression patterns indicate that larval feeding decisions involve dynamic central integration rather than simple reflexive responses. The CNS likely integrates feeding-promoting signals (such as sNPF, NPF, SIFa, and AKH) with peripheral gustatory inputs to form a unified representation of hunger and food quality [61,62,63,64]. In contrast, insulin-like peptide signaling monitors internal energy status [65] and may link metabolic demand with sensory perception through insulinergic modulation of neural activity [66,67]. Together, these signals constitute a coordinated network capable of fine-tuning feeding behavior in response to internal states and external cues.

Interestingly, the CNS also expresses detoxification-related genes (Table S6), implying that its function extends beyond nutrient sensing and may involve the preliminary recognition and evaluation of such compounds [68,69,70,71]. The interaction, and potential antagonism, between these “feeding-promoting” and “feeding-avoidance” signals may establish a dynamic equilibrium at the neural circuit level, thereby regulating the initiation, maintenance, or termination of feeding behavior [72,73,74]. This integrated model elucidates the molecular and systems-level basis of larval dietary plasticity, revealing that feeding behavior results from the CNS simultaneously integrating external chemical signals with internal physiological states.

Based on these expression profile characteristics, we propose a multi-level neuromolecular coordination model for larval feeding behavior. This model posits that when larvae encounter host plants, nutrients such as sugars and amino acids activate specific peripheral taste receptors and sensory neurons. This activation is likely to trigger a conserved G protein—IP3 signaling cascade in the CNS—a pathway that may share similarities with pheromone processing in locusts [75]. This cascade rapidly amplifies the initial chemical signal and converts it into an intracellular response, generating a core motivational signal that drives the initiation of feeding [76,77].

In this study, differential expression, functional enrichment, and evolutionary relationship analyses were performed to identify and prioritize key candidate genes, which were subsequently validated via qPCR. Given their distinct functional roles, the brain is regarded as the high-level integration center likely responsible for feeding motivation and decision-making, whereas the GNG primarily processes peripheral gustatory signals and coordinates feeding movements. Therefore, by systematically comparing these two tissues in terms of gene expression, signaling pathways, and neuropeptide regulatory networks, this study aims to elucidate how different regions of the CNS collaboratively regulate feeding behavior. Collectively, this integrated strategy not only provides a molecular basis for understanding the organizational principles of insect feeding neural circuits but also offers potential insights for developing targeted pest control strategies.

5. Conclusions

Collectively, this study provides the first systematic characterization of feeding-mediated genes in the gustatory sensory centers—the brain and GNG—of the H. armigera larvae. Key findings include the identification of 944 differentially expressed genes, with 41 of these determined as key candidate genes associated with feeding behaviors of H. armigera larvae, including neuropeptides, neuropeptide receptors, gustatory receptors, and feeding-mediated proteins. These genes are evolutionarily conserved, exhibit elevated expression in the brain and GNG of fifth-instar larvae, and are significantly upregulated under starvation conditions. These results advance our understanding of the molecular mechanisms that regulate insect feeding and suggest a foundation for developing sustainable pest control strategies that target gustatory or neuropeptide signaling in this polyphagous pest species.