Evolutionary Novelty of Apolipoprotein D Facilitates Metabolic Plasticity in Lepidopteran Wings Authors: Shunze Jia, Rongqiao Li, Yinghui Li, Yuxin Huang, Minmin Liu, Yanyan Zhou, Yanting Liang, Zhihua Hao, Yusong Xu, Huabing Wang Published: 2024 ABSTRACT AbstractUnderstanding metabolic plasticity of animal evolution is a fundamental challenge in evolutionary biology. Owing to the diversification of insect wing morphology and dynamic energy requirements, the molecular adaptation mechanisms underlying the metabolic pathways in wing evolution remain largely unknown. This study reveals the pivotal role of the duplicated Apolipoprotein D (ApoD) gene in lipid and energy homeostasis in the lepidopteran wing. ApoD underwent significant expansion in insects, with gene duplication and consistent retention observed in Lepidoptera. Notably, duplicated ApoD2 was highly expressed in lepidopteran wings and encoded a unique C-terminal tail, conferring distinct ligand-binding properties. Using Bombyx mori as a model organism, we integrated evolutionary analysis, multiomics, and in vivo functional experiments to elucidate the way duplicated ApoD2 mediates lipid trafficking and homeostasis via the AMP-activated protein kinase pathway in wings. Moreover, we revealed the specific expression and functional divergence of duplicated ApoD as a key mechanism regulating lipid homeostasis in the lepidopteran wing. These findings highlight an evolutionary scenario in which neofunctionalization conferred a novel role of ApoD in shaping adaptive lipid metabolic regulatory networks during wing phenotypic evolution. Overall, we provide in vivo evidence for the functional differentiation of duplicate genes in shaping adaptive metabolic regulatory networks during phenotypic evolution. Introduction Metabolic flexibility is the ability to adapt metabolism efficiently based on demand or supply of energy resources and requirements (Smith et al. 2018; Koyama et al. 2020). The evolution of metabolic pathways represents a crucial step in molecular and cellular evolution. Various factors have influenced the formation of genomes and the development of novel metabolic capabilities in animals. Population genetic studies have indicated that selection based on core metabolic enzymes and transport proteins can shape life histories by modifying metabolic flux (Williams et al. 2016). However, the regulatory mechanisms underlying metabolic flexibility remain largely unknown. Despite descriptions of the functional morphology of wings in insects, the relationship between evolutionary driving forces and adaptive metabolic processes has only been comprehensively explored in a few studies (Tsai et al. 2020). Wing evolution has enabled insects to become one of the most diverse clades on Earth. Although winged insects typically possess two pairs of wings, substantial structural, functional, and morphological differences exist among species. The evolution of winged insects (Pterygota) from primitive wingless forms marked a seminal development in insect body design (Wootton and KukalovÃ-Peck 2000; Medved et al. 2015; Stork et al. 2015). In Neoptera, the ability to fold wings over the body facilitated evasion of wing damage while maneuvering through tight spaces (Engel 2015). Additionally, the evolution of insect wings involves essential lipid and energy metabolism pathways required to meet diverse energy demands. Lipid metabolism and homeostasis play crucial roles in growth and patterning during insect wing development (Panáková et al. 2005). For example, lipid trafficking is critical for insect wing growth and patterning through Hedgehog and Wingless signaling (Panáková et al. 2005). Intensive energy expenditure and membrane lipid synthesis occur during extensive cell growth and proliferation in wing imaginal discs during the larval and pupal stages (Fauny et al. 2005; Palm et al. 2012). Additionally, given their high metabolic rate and elaborate morphology, insect wings require a continuous supply of lipids. Owing to the diversity of lipid transport proteins and complexity of their tissue- and cell-specific expression, the adaptation mechanisms underlying the wing lipid metabolism system remain poorly understood. Animal lipid transport processes rely on specialized lipoprotein complexes, including apolipoproteins that stabilize lipid components and mediate lipid metabolism. Apolipoprotein D (ApoD) is an ancient, multiligand lipoprotein that binds various lipophilic ligands such as cholesterol, steroids, and retinoic acid (RA) (Rassart et al. 2000; Dassati et al. 2014). ApoD is a multifunctional protein involved in lipid trafficking, lipid metabolism regulation, longevity, stress resistance, and antioxidative response (Rassart et al. 2000; Sanchez et al. 2006; Dassati et al. 2014). ApoD expression in metabolic tissues of insects is associated with insulin pathway and glucose homeostasis in a tissue-dependent manner. Drosophila melanogaster has two ApoD paralogs: Glial Lazarillo (Glaz) (Sanchez et al. 2006; Walker et al. 2006) and Neural Lazarillo (Nlaz) (Hull-Thompson et al. 2009; Ruiz, Sanchez et al. 2013). Loss of Glaz or Nlaz reduces resistance to oxidative stress and diminishes fat storage (Sanchez et al. 2006; Walker et al. 2006; Hull-Thompson et al. 2009). GLaz mediates neuron–glia lipid shuttling and supports dendrite morphogenesis in Drosophila (Yin et al. 2021). Infection-induced unpaired cytokine expression activates JAK–STAT signaling in ensheathing glia, inducing GLaz expression and modulating metabolic coupling of glia and neurons at the antennal lobe of Drosophila (Cai 2021). Unlike the mammalian ApoD, which has only one copy (Rassart et al. 2000), insect ApoD homologs appear to have undergone gene duplication and functional divergence. In previous studies, we revealed the duplication events of the ApoD gene in several insect orders (Zhou et al. 2018, 2020). Lepidoptera, the second largest order of insects with more than 100,000 described species, exhibits a wide range of ecological adaptations (Misof et al. 2014; Kumar et al. 2017). Despite similarities to other insects in many aspects of their biology, lepidopteran species have evolved numerous novel traits through analogous ecological life strategies. Our previous research revealed that three lepidopteran ApoD paralogs expanded universally via gene duplication, with ApoD1 and ApoD3 being orthologous to Glaz and Nlaz in D. melanogaster, respectively (Zhou et al. 2018, 2020). Interestingly, our transcriptional data demonstrated that the duplicated ApoD2 was the most highly expressed ApoD paralog in lepidopteran wings. Therefore, exploring the role of ApoD2 in wing lipid transport and supply adds an intriguing dimension to the study of ApoD duplications. In this study, we conducted a phylogenetic analysis involving 100 insect species in Apterygota and Pterygota to reconstruct the evolutionary relationships of ApoD. Duplication events of ApoD genes were observed in several insects and extensively preserved in lepidopteran insect genomes. We deleted ApoD2 or ApoD3 in a lepidopteran model insect, Bombyx mori. Notably, the forewings and hindwings of duplicated ApoD2 mutants exhibited significantly reduced wing surface area in the adult stage. Comprehensive multiomics and functional assays were conducted to investigate the molecular mechanisms underlying physiological functions of ApoD2 in the wing. Our results highlight the specific expression and functional divergence of duplicated ApoD, which plays a key role in mediating lipid shuttling in lepidopteran wings. The present study provides insights into the contribution of gene duplication to regulatory network architecture and metabolic adaptation, improving our understanding of the genetic mechanisms underlying metabolic plasticity and driving evolutionary innovation. Results Origin of the ApoD Gene Family in Insects ApoD plays critical roles in longevity, stress resistance, and lipid metabolism regulation. To characterize systematically the evolutionary history of ApoD genes in insects, we performed HMMER and Blastp searches across 100 insect species to identify putative ApoD orthologous genes. We identified 231 putative ApoD genes in the genomes of 100 insects (supplementary table S1, Supplementary Material online). Genomes of most Diptera and Hymenoptera exhibit one or two ApoD homologs. Notably, we observed three or four ApoD paralogs in lepidopteran insects (supplementary fig. S1, Supplementary Material online). The analysis indicated a variable number of ApoD genes among these species. To test evolutionary history of the paralogs and the three types of ApoD genes in insects rigorously, we conducted a phylogenetic study using the maximum likelihood (ML) method (Fig. 1). The phylogenetic analysis identified three subfamily genes, for which the gene branching order indicated that ApoD paralogs are monophyletic, with ApoD2 and ApoD3 derived from ApoD1. Interestingly, ApoD1 and ApoD2 were found in Coleoptera, whereas ApoD3 was missing in most coleopteran insects. By contrast, ApoD2 and ApoD3 were found in most hymenopteran insects, whereas ApoD2 was lost in most dipteran insects. Notably, ApoD1, ApoD2, and ApoD3 were almost completely retained in 21 lepidopteran species. Phylogenetic analysis of ApoD genes in insects. ML tree was constructed using RAxML software based on the (LG + I + G) model. The best substitution models were determined using ProtTest. All amino acid sequences from genomes of 100 species were acquired using HMMER and Blastp. Despite their essential role in lipid metabolism, ApoD genes showed evolutionary lability, with newly duplicated ApoD genes absent from multiple insect lineages (Fig. 1). Our investigation into the location, order, and orientation of ApoD genes revealed loci for ApoD2 and ApoD3 were syntenic and conserved across all lepidopteran species (Fig. 2). Moreover, their order and orientation were highly conserved. Therefore, our syntenic and phylogenetic analyses highlight the prevalence of conserved ApoD2 gene duplications in lepidopteran species. Conserved syntenic analysis of ApoDs and their surrounding genes. Synteny was analyzed using ApoD1, ApoD2, and ApoD3 as anchor sites. Homologs are represented by the same color, whereas genes with no homolog are indicated by white boxes. ApoD genes are shown in red. An Extensive C-terminal Tail Allowed ApoD2 to Have a Distinct Ligand-Binding Property Duplicated genes associated with distinct protein motifs and evolutionary constraints may undergo functional divergence (Kursel and Malik 2017; Kursel et al. 2020). We investigated the primary structure of lepidopteran ApoD proteins (supplementary figs. S2 and S3, Supplementary Material online) and observed that in ApoD2, three structurally conserved regions of lipocalins were highly conserved, whereas the C-terminal tails exhibited divergence. The unique C-terminal tail (∼100 amino acids) of lepidopteran ApoD2 extended from the conserved motifs shared with other ApoDs. The acquisition of this novel C-terminal tail by lepidopteran ApoD2 relative to ApoD1 and ApoD3 implied that lepidopteran ApoD2 may perform at least a partially nonredundant function. In line with this hypothesis, our previous research demonstrated positive selection has occurred in lepidopteran ApoD2, potentially causing conformational changes that could influence its binding properties (Zhou et al. 2020). These results suggest that the divergent evolution of the C-terminal tails may have conferred distinct ligand-binding abilities. To explore whether the divergent C-terminal tail influences specific binding properties, we performed a ligand-binding test. We expressed and purified recombinant ApoD2, ApoD3, and a C-terminal tail-deleted form of ApoD2 (Trunc-ApoD2) from the lepidopteran model species B. mori (Fig. 3a and supplementary fig. S4, Supplementary Material online). We assessed ligand-binding properties using small hydrophobic molecules, including retinoic acid (RA), 20-hydroxyecdysone (20E), and ergosterol (supplementary fig. S5, Supplementary Material online). 20E plays important roles in larval molting and metamorphosis (Riddiford et al. 2000). Ergosterol is the major sterol in several insect foods (Rietveld et al. 1999). Although RA is a common ligand for BmApoD2 and BmApoD3, neither BmApoD2 nor BmApoD3 exhibited positive binding to 20E. Based on surface plasmon resonance (SPR) analysis, BmApoD2 could bind more tightly to ergosterol than BmApoD1 and BmApoD3 (Fig. 3b and supplementary fig. S6, Supplementary Material online). The equilibrium dissociation constant (KD) of Trunc-ApoD2 binding to ergosterol decreased by ∼268-fold from 1.61 × 10−9 M to 4.33 × 10−7 M. Deletion of the C terminus significantly decreased the affinity of BmApoD2 to ergosterol, which indicates that the increase in binding affinity for this complex is primarily driven by the C-terminal tail. These results suggest that the extensive evolving C-terminal tail endows ApoD2 with distinct ligand-binding properties. Ligand-binding and expression pattern of ApoDs. a) The gene structure and protein domains of BmApoDs are displayed. The numbers above the gene structure indicate gene localization, with an arrow pointing to the ATG start codon, while the white boxes represent the untranslated regions. The protein structure is shown below the gene structure, with the numbers above indicating the amino acid positions. The common domain (Lipocalin_apoD-like) is highlighted in the figure. Trunc-BmApoD2 is a truncated form of BmApoD2 that lacks 74 amino acids in the C-terminal tail. b) SPR binding affinity measurement of B. mori ApoDs and Ergosterol. Recombinant BmApoD1, BmApoD2, BmApoD3, or Trunc-BmApoD2 protein was immobilized on the sensor chip (CM5) using the amine-coupling method. The data were collected using multicycle kinetics. c) ApoD expression patterns in different tissues and stages of B. mori. Quantitative real-time PCR estimates of relative BmApoD1, BmApoD2, and BmApoD3 mRNA levels. Hd, head; Ep, epidermis; Wg, wing disc; Mg, midgut; ASg, anterior silk gland; MSg, middle silk gland; PSg, posterior silk gland; fFb, female fat body; mFb, male fat body; Mt, malpighian tubules; Te, testis; Ov, ovary. Data are shown as mean ± SEM, n = 3. P values were calculated by one-way ANOVA. **, P < 0.01; ****, P < 0.0001. Expression Divergence of Duplicated ApoD in Lepidopteran Insects Gene expression levels play a pivotal role in understanding the dynamics of functional divergence and the emergence of evolutionary novelty (Guschanski et al. 2017). To evaluate the expression divergence of ApoD genes, we examined two lepidopteran model insects, Papilio xuthus (a butterfly species in Papilionidae) and B. mori (a moth species in Bombycidae). We observed spatial and temporal expression profiles of ApoD in diverse tissues at three key developmental stages: day 3 of the fifth larval instar (L5D3), day 3 of the pupa (P3), and day 1 of the adult (A1). In B. mori, BmApoD2 showed the highest expression levels in the wing disc at L5D3, which further increased at the P3 stage (Fig. 3c). At P3, a critical stage for wing development in holometabolous insects (Truman 2019), the expression levels of BmApoD2 in the wings were significantly higher than in other tissues. BmApoD1 was expressed in multiple tissues, with notable enrichment in the testis at the A1 stage, while BmApoD3 showed high expression in the head at the L5D3 stage and limited expression in the wings at the P3 stage. These findings underscored the diverse and dynamic expression of ApoD genes across different tissues and stages. Notably, BmApoD2 showed the highest expression in wings, a pattern further confirmed by consistent protein levels observed in western blot analysis (supplementary figs. S7 and S8a, Supplementary Material online). Further analysis revealed BmApoD2 protein expression in wing development from the middle-to-late larval stages, a sharp decline during wandering, and predominant enrichment in early pupae, with no expression during nonfeeding (wandering and spinning) stages (supplementary fig. S8b, Supplementary Material online). Given that 20E promotes wing disc differentiation and development in vivo and in vitro (Fujiwara and Hojyo 1997; Obara et al. 2002) and its levels increase for ∼20 h during the last larval instar (Bollenbacher et al. 1981), we investigated whether BmApoD2 expression was regulated by 20E. In vitro cultivation of B. mori wing discs at L5D3, with or without 20E for 6 h, resulted in dose-dependent inhibition of BmApoD2 expression in the presence of 20E (supplementary fig. S8c, Supplementary Material online). In P. xuthus, high ApoD2 expression levels were also found in wings (supplementary fig. S9, Supplementary Material online). The prevalence of elevated ApoD2 expression in this tissue compared with other ApoD genes suggests its possible association with wing development. Collectively, these results demonstrate that ApoD2 was primarily expressed in the wings of lepidopteran insects. ApoD2 Impacts Lipid Metabolism in B. mori Wings To investigate the physiological functions of ApoD2 and ApoD3 genes comprehensively, we used the CRISPR/Cas9 system to generate specific knockout silkworms for BmApoD2 and BmApoD3. We obtained two major BmApoD2 mutant lines featuring 5- or 6-bp deletions and a BmApoD3 mutant line with a 5-bp deletion. All homozygous silkworms remained viable (supplementary fig. S10, Supplementary Material online). Western blot analysis confirmed the complete absence of the BmApoD2 protein in the wings of BmApoD2 mutants at all three stages (Fig. 4a). At the pupal stage, our immunofluorescence staining further revealed that the deletion of BmApoD2 reduced its expression (Fig. 4b). Interestingly, the BmApoD2 mutant of the pupal stage displayed a deeper coloration, with a notable deposition of a melanin analog on the wings (Fig. 4c). Importantly, both male and female forewing and hindwing exhibited significantly reduced surface areas in the adult stage (Fig. 4d). In contrast, the wings of the BmApoD3 mutant showed no significant differences compared with the wild type. All three 5-bp BmApoD2−/− deletions displayed the same phenotype. Therefore, we selected the silkworm line carrying the initial 5-bp deletion (ACCCG) of BmApoD2−/− for subsequent experiments. The deletion of BmApoD2 causes defects in wing development. a) Protein levels were detected at different stages in the wings of wild-type and mutant silkworms. b) Representative images of immunofluorescence from wing discs of silkworms with or without BmApoD2 at the P3 stage. The BmApoD2 channel was quantified by relative intensity using ImageJ. Scale bar: 1 mm. c) The representative images show the wing color of wild-type and BmApoD2−/− silkworm pupae at P3, with each genotype having 10 replicates (five males in the first row and five females in the second row). d) Wings and wing area of wild type and BmApoD2 mutants for both sexes at the adult stage (50 males and 50 females per group with three groups). Data are shown as mean ± SEM. P values were calculated by unpaired t-test or one-way ANOVA. *P < 0.05; ****P < 0.0001. BmApoD2 Mediates Lipid Transport in Wings During the Pupal Stage To gain insights into the function of BmApoD2 in the wings of silkworms, we performed iTRAQ-based quantitative proteome analysis of wings from both the wild type and BmApoD2 mutants at the P3 stage. iTRAQ quantification identified 327 differentially expressed proteins (DEPs) comprising 177 upregulated and 150 downregulated DEPs (fold-change ≥ 1.2 or ≤ 0.8333; P < 0.05; supplementary table S2, Supplementary Material online). Clusters of Orthologous Groups of Proteins (COG) analysis of all DEPs revealed nearly twice as many upregulated proteins related to energy production compared with downregulated ones (Fig. 5a and supplementary fig. S11a, Supplementary Material online), indicating a significant increase in energy demand at the early pupal stage. Consistent with the BmApoD2 deletion phenotype, most DEPs involved in lipid transport and metabolism were downregulated in the wings of BmApoD2 mutants. BmApoD2 mediates lipid transport and lipid metabolism homeostasis. a) COG functional analysis of DEPs in the wild type and BmApoD2−/− mutants at the P3 stage. The lollipop graph on the lower axis indicates the number of proteins upregulated or downregulated in the corresponding pathways, while the scatter plot on the upper axis shows the related fold changes of the proteins. b) Enrichment pathway analysis of DEMs. c) Heatmap of 79 DEMs annotated as lipids and lipid-like molecules. Each column in the left or right panel represents one replicate. d) The diagram of the linoleic acid metabolism pathway, with red circles indicating metabolites that are upregulated in the mutant. e) The box plot showing the levels of selected metabolites. d, e) Data representing the ratio of the average expression of metabolites in the wild type and BmApoD2−/− mutants at P3 (log2FC) (*P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001). Detailed annotations of DEMs are provided in supplementary table S3, Supplementary Material online. To further identify potential metabolites associated with BmApoD2, we performed metabolome analysis in wings at P3. A partial least squares-discriminant analysis (PLS-DA) score plot of the GC–MS metabolome showed significant differences in metabolic characteristics between the control and BmApoD2 mutants at P3 (supplementary table S3, Supplementary Material online). The principal components between the control and BmApoD2 mutants in positive mode (17.72 and 25.74%, respectively) and PC1 and PC2 in negative mode (20.14 and 33.69%, respectively) were clearly separated (supplementary fig. S11b, Supplementary Material online). Simultaneously, we detected 615 and 320 differentially expressed metabolites (DEMs; variable importance on projection [VIP] ≥ 1, fold-change ≥ 1.2 or ≤ 0.8333, and P < 0.05) in the positive and negative models, respectively. Among these 935 DEMs, lipids and lipid-like molecules (79 DEMs), benzenoids (24 DEMs), and organic oxygen compounds (16 DEMs) were the most highly annotated DEMs based on super classification. Mapping the DEMs to those in the Kyoto Encyclopedia of Genes and Genomes (KEGG) metabolic pathway database showed that the pathways with significant DEM enrichment were mainly involved in starch and sucrose metabolism, galactose metabolism, linoleic acid metabolism, biosynthesis of unsaturated fatty acids, and riboflavin metabolism (Fig. 5b). Importantly, 62 DEMs annotated with lipids and lipid-like molecules were downregulated upon BmApoD2 disruption (Fig. 5c). Metabolome analysis indicated that BmApoD2 mutant wings contained significantly more JH I and JH III. Juvenile hormone (JH), a key player in various biological processes in insects, showed significant differences in its chemical forms. The most common form was JH III, although lepidopterans also produce homologous JH structures, including JH 0, JH I, methyl-JH I, and JH II. SPR analysis showed that BmApoD2 can interact with JH III, although the interaction seems to be weak (supplementary fig. S12, Supplementary Material online). These results suggest BmApoD2 is involved in the transport of JH. Insects typically have a higher content of unsaturated fatty acids, including linoleic and linolenic acids (Fig. 5d and e and supplementary table S3, Supplementary Material online). These unsaturated fatty acids play a crucial role in energy supply during nonfeeding stages (Arrese and Soulages 2010). Interestingly, the linoleic acid metabolism pathway in the wings of BmApoD2 mutants was significantly upregulated compared with the wild type (Fig. 5d and e), suggesting that the absence of BmApoD2 leads to an imbalance in linoleic acid metabolites. Similarly, our RNA-seq analysis found that BmApoD2 participates in metabolic stress sensing and in the lipid metabolic network (supplementary fig. S13, Supplementary Material online). To gain insights into BmApoD2 function at both larval and pupal stages, we performed RNA-seq analysis of wings from the wild type and BmApoD2 mutants at L5D3 and P3 (supplementary tables S4 and S5, Supplementary Material online). In gene set enrichment analysis (GSEA), typical stimulation-induced Gene Ontology (GO) terms were associated with L5D3, including innate immune response, stimulating response, and extracellular region (supplementary fig. S13, Supplementary Material online). Functional analysis using KEGG pathways revealed significant enrichment in metabolism-related processes (supplementary fig. S13, Supplementary Material online). For instance, genes related to glucose metabolism and phagosome pathways were overrepresented. DEGs enriched in fatty acid metabolism and the fatty acid elongation pathway varied in direction at the pupal stage, indicating that BmApoD2 affects lipid and fatty acid metabolism in wings during this stage. Taken together, these findings underscore the crucial role of BmApoD2 in maintaining hormonal homeostasis and regulating specific metabolites in the wing. BmApoD2 Facilitates Lipid Homeostasis in the Wing ApoD plays a well-established role in lipid transport and metabolism. Metabolome analysis highlighted significant lipid and lipid-like changes in the wings of BmApoD2 mutants. Therefore, we further conducted lipidomics analysis to compare wing lipid composition between the wild type and BmApoD2 mutants, which identified 354 lipids and their comprehensive annotations (supplementary table S6, Supplementary Material online). Loss of BmApoD2 triggered substantial changes in wing disc lipids, yielding 106 differentially regulated lipid molecules (VIP ≥ 1, fold-change ≥ 1.2 or ≤ 0.8333, and P < 0.05) (supplementary fig. S14, Supplementary Material online), which comprised 19 subclasses containing 23 phosphatidylethanolamines, 21 phosphatidylcholines, 12 sphingomyelins, 10 phosphatidylinositols, 8 ceramides, 6 diacylglycerols, 4 triacylglycerols, 4 phosphatidylserines, 2 phosphatidylglycerols, and 2 lysophosphatidylcholines, as well as other molecules (Fig. 6a and supplementary fig. S14, Supplementary Material online). Diacylglycerols are the main lipids in the hemolymph of several insects (Canavoso et al. 2001), including phosphatidylethanolamines, triacylglycerols, phosphatidylcholines, and ceramides. Interestingly, 6 diacylglycerols and 17 phosphatidylethanolamines were significantly downregulated after BmApoD2 disruption. Conversely, 10 sphingomyelins, 8 phosphatidylinositols, and 8 ceramides were significantly upregulated, suggesting that the same type of lipid molecules showed similar trends. Loss of BmApoD2 impairs lipid and energy metabolism. a) Lipid changes in wings from the wild type and BmApoD2−/− mutants at P3, indicating significant upregulated and downregulated features detected via global lipid profiling. Lipid subclasses include triglyceride (Tg), diglyceride (Dg), phosphatidylcholine (Pc), phosphatidylethanolamine (Pe), ceramide (Cer), phosphatidylinositol (Pi), phosphatidylserine (Ps), phosphatidylglycerol (Pg), sphingomyelin (Sm), and lysophosphatidylcholine (Lpc). Lipids exhibiting a consistent trend are indicated using colored fonts. Details of the detected and characterized lipids are provided in supplementary table S6, Supplementary Material online. b) Nile red staining of lipid droplets in P3 wings from different genetic backgrounds. Wild-type wings served as a control. BmApoD2 loss induced a lipid storage defect in the wings. Nuclei were stained with DAPI (blue). Scale bar: 100 µm. (c, d) Biochemical influence of electron transport chain components following BmApoD2 deletion (*P < 0.05; **P < 0.01). e) Developmental phosphorylation and protein profile of AMPK in silkworm wings from the pupal stage. Consistent with the multiomics data, we found BmApoD2 disruption led to a lipid storage defect in P3 wings (Fig. 6b). We were not surprised to find reduced complex I activity and ATP levels in the wings of BmApoD2 mutant pupae (Fig. 6c), which is consistent with the established association of lipid metabolism with energy homeostasis (Arrese and Soulages 2010). The AMP-activated protein kinase (AMPK) pathway is crucial for reprogramming adaptive metabolism under low-energy conditions (Mihaylova and Shaw 2011). To confirm the role of BmApoD2 in energy metabolism, we compared AMPKα phosphorylation profiles between the wild type and BmApoD2 mutant. Examining AMPKα phosphorylation levels at Thr172 in wings during the entire pupal stage, we found increased phosphorylation in the wings of BmApoD2 mutants during the early pupal stage compared with the controls (Fig. 6d). This stage-specific trend highlights the role of AMPKα activation in BmApoD2-mediated energy metabolism. Taken together, these results suggest that BmApoD2 delivers lipids to wings for energy metabolism and provides membrane phospholipids to wing discs, highlighting a potential conserved feature of wing lipid metabolism through ApoD2. Discussion The evolution of wings is a critical aspect of insect history. Owing to the diversity of lipid transport proteins and the complexity of their tissue- and cell-specific expression, the evolutionary trajectory of lipid metabolic pathways in insect wings remains unclear. The present study highlights the vital role of the duplicated ApoD2 gene in coordinating the regulation of lipid metabolism and metabolic homeostasis in insect wing development. Employing CRISPR/Cas9-mediated genome editing with lipidomics analysis, we revealed unexpected lipid transport pathways mediated through disruption of BmApoD2, which markedly affected lipid metabolic homeostasis in these lepidopteran wings. Our multilevel investigation indicated that duplication of ApoD facilitated lepidopteran wing metabolic adaptability. Consequently, this study provides insights into the contribution of duplicated genes to metabolic adaptation and expands our understanding of wing phenotypic evolution and insect diversity. In insects, lipid transport primarily relies on major hemolymphatic lipoproteins, which facilitate the distribution of diverse lipid classes across storage, absorption, and utilization sites. ApoD is a multifunctional protein, participating in regulation of lipid metabolism, antioxidative response, and stress resistance (Kalman et al. 2000; Rassart et al. 2000; Dassati et al. 2014). Our study revealed the genomic landscape of duplicated ApoD in insects. Although most insects possess two ApoD genes, certain species encode an additional ApoD. ApoD1 genes across all insects form a monophyletic clade, whereas ApoD2 and ApoD3 genes constitute monophyletic sister clades, indicating potential evolution through a duplication event. Notably, ApoD duplication is almost completely preserved in lepidopteran species. Our synteny-based findings support the results of the phylogenetic analysis, suggesting that ApoD2 and ApoD3 are tandemly duplicated and consistently preserved in their shared syntenic locations across all lepidopteran species examined, which represent 10 major clades. The C-terminal tails of ApoD2 and ApoD3 appear to have undergone divergent evolution. Unlike the evolutionarily conserved N-terminal domain, the C-terminal tail of ApoD2 displayed high variability in length and sequence. Our previous work revealed distinct evolutionary rates among ApoD paralogs, as indicated by variations in dN/dS ratios (Zhou et al. 2020). Positive selection was detected in the lepidopteran ApoD2 lineage, with positively selected sites localized around the pocket and loop domains, suggesting conformational changes that may influence ligand-binding properties (Zhou et al. 2020). Consistent with this, our ligand-binding experiments demonstrated that ApoD2 molecules exhibit both shared and unique functionalities in ligand transport. This supports the notion that duplicated ApoD2 likely undergoes partial neofunctionalization, an evolutionary trajectory akin to paralogous mammalian antiviral proteins, IFIT1 and IFIT1B, wherein the divergent C terminus confers distinct antiviral specificity (Daugherty et al. 2016). However, this issue remains to be resolved because it is possible that duplicated ApoD genes undergo rapid subfunctionalization, followed by a prolonged period of neofunctionalization. The contribution of duplicated genes to the emergence of novel functions may be manifested in the tissue-specific expression of paralogs originating at distinct evolutionary times and later acquiring divergent sequences (Zhang et al. 2015). Evaluating the fates of conserved lepidopteran ApoD2, we also assessed the divergence of ApoD paralog expression. ApoD2 exhibited high expression in the wings of both B. mori and P. xuthus, highlighting their potential contribution to lepidopteran wing development. Insect wing discs use lipids as cellular membranes, signaling molecules, and major energy stores. In the present study, the disruption of BmApoD2 affected lipid metabolism including JH levels at the pupal stage and resulted in a reduced wing area phenotype. Untargeted metabolomics data revealed that BmApoD2 interruption exerted pleiotropic effects on several metabolic pathways, with a higher number of downregulated lipids and lipid-like molecules compared with their upregulated counterparts. Insects continually expend energy, relying on accumulated reserves during nonfeeding stages (Arrese and Soulages 2010). Lipids are the primary fuel during holometabolous metamorphosis (Merkey et al. 2011), with insect lipids characterized by a high unsaturated fatty acid content, exceeding 2.5 times that of saturated fatty acids in certain insects (Nakasone and Ito 1967; Raksakantong et al. 2010). Network analysis revealed consistent upregulation of most metabolites in the linoleic acid metabolism pathway in the wings of BmApoD2 mutants. Furthermore, multiomics analysis showed that significantly altered metabolites were associated with various metabolic pathways. BmApoD2 delivers these lipids to the wings, where they may be used for energy metabolism. Additionally, BmApoD2 contributes a significant fraction of membrane phospholipids to wing discs. JH, characterized by a unique terpenoid structure, is the methyl ester of epoxy farnesoic acid. JH III is the predominant form of JH in most insects, although JH I and II are the principal JHs in lepidopteran insects. Notably, JH I and III were significantly upregulated in the wings of BmApoD2 mutants, indicating that BmApoD2 modulates wing JH levels to maintain hormone homeostasis. Uncovering the precise roles of these metabolites in wing development is an essential subject for future research. To explore the impact of ApoD2 on wing lipid composition, we performed untargeted lipidomics profiling. We identified ApoD2-dependent lipid mobilization, which contributed to the neutral and polar lipid composition of wing imaginal discs. For example, metabolites of the linoleic acid metabolism pathway were mainly upregulated in the wings of BmApoD2 mutants. These findings delineate interorgan lipid transport routes through ApoD2 in the wing disc of B. mori, revealing unexpected wing-specific variations in lipid utilization. ApoD2 transports lipids to wing imaginal discs, fostering the development of their fat stores and enhancing energy use, and contributes a major fraction of membrane phospholipids to wing discs. Previous studies have shown that lipoprotein particles are required for Hedgehog and Wingless signaling in the developing wings of Drosophila (Panáková et al. 2005). In lipophorin–RNAi larvae, the wing disc exhibits reduced size, particularly in the wing pouch, compared with the wild type (Panáková et al. 2005), as Hedgehog fails to signal over its normal range and the Wingless signaling range is narrowed. A recent study found that several ApoD-like genes with distinct spatiotemporal expression patterns were found in Nilaparvata lugens, whereas functional screening using an RNA interference approach revealed that only ApoD2 was essential for development and survival (Lu et al. 2024). Future investigations leveraging ApoD2 to modulate JH levels and Wingless pathway activity will enhance our understanding of the molecular mechanisms connecting nutrition, signaling, and developmental growth and patterning. At the early pupal stage, BmApoD2 disruption led to poor energy status and increased AMPK phosphorylation in the wings. AMPK can phosphorylate several key transcription factors that serve as master regulators of biosynthetic pathways, metabolism, and autophagy (Herzig and Shaw 2018). In response to a declining energy status, the activated AMPK pathway may reprogram lipid metabolism in the wings of BmApoD2 mutants. These results emphasize the vital role played by ApoD2 in determining JH levels and lipid metabolism in lepidopteran wings. Our study delineates a fundamental aspect of wing lipid metabolism, highlighting novel mechanisms by which duplicated ApoD2 affects lipid metabolism more broadly (Fig. 7). Furthermore, our findings support the hypothesis that the duplication and retention of ApoD2 reflect the diversification and varied demand for lipid metabolism in lepidopteran wings. Studying the evolution of metabolic pathways offers valuable insights into metabolic flexibility, and we speculate that ApoD2 confers metabolic plasticity to lepidopteran wings, thereby promoting metabolic adaptability. Evolutionary novelty of ApoD contributes to lipid homeostasis and energy balance in lepidopteran wings. a) Summary of the evolutionary novelty and history of ApoD in Lepidoptera. The black ApoD represents the ancestral ApoD. ApoD2 and ApoD3 are generally conserved in Lepidoptera. The unique C terminus of ApoD2 confers a new function in wing development. b) Summary of the novel function and expression pattern of ApoD2 in lepidopteran wings. The expression levels of ApoD2 are indicated by a gradient from dark red (representing higher expression in the bars on the top) to light red (representing lower expression in the bars on the bottom). The triangle summarizes the role of ApoD2 in the wing. In summary, we systematically and quantitatively explored how duplicated ApoD2 influences lepidopteran wing lipid composition. By integrating evolutionary analysis, multiomics analysis, and functional assays, we demonstrated the crucial role of BmApoD2 in maintaining lipid and energy homeostasis during wing development. We conclude that duplicated ApoD2 is indispensable not only for hydrophobic ligand transport but also for coordinate regulation of lipid and energy metabolism, which is vital for wing development. Overall, these findings emphasize the role of duplicated ApoD2 in facilitating metabolic adaptation of lepidopteran wings, offering insights into the genetic mechanisms underlying organismal adaptation and evolutionary innovation in insects and eukaryotes more broadly. Materials and Methods Phylogenetic and Evolutionary Analyses A hidden Markov model profile was constructed from multiple sequence alignment results using HMMER v3.3.2 (http://hmmer.org). HMMER hits with an e-value of ≤1e−30 were validated for accuracy, and partial hits were excluded (Hering et al. 2012). Multiple amino acid sequence alignment analyses were performed using the MAFFT program. ML analyses were performed using IQTree v.1.5.5, with automatic model selection by the software (Nguyen et al. 2015). Visualization of phylogenetic trees was accomplished using ggtree (Yu et al. 2018) and iTOL (Letunic and Bork 2019). Construction of Cas9 and Single Guide RNA Expression Systems Bombyx mori (N4 strain) served as the wild-type control. Mutant alleles for BmApoD2 were generated using the CRISPR/Cas9 system (Hwang et al. 2013; Mali et al. 2013). Cas9 mRNA was transcribed using the mMESSAGE mMACHINE T7 Kit (Ambion, USA). Independent guide RNAs (gRNAs) were designed using a gRNA design tool (https://crispor.gi.ucsc.edu/). Sequences are listed in Supporting Information (supplementary table S7, Supplementary Material online). Transcription experiments were conducted using the MEGAscript T7 Kit. A mixture with a 10 μL volume, containing 1,000 ng of single guide RNA and 3,000 ng of Cas9 mRNA, was injected into embryos within 6 h after oviposition. Homozygote Screening Genomic DNA from viable silkworms was extracted from the dissected wings of moths using a Genome DNA Isolation Kit (Sangon Biotech, China). The PCR-amplified products containing the target site were sequenced directly. Offspring from heterozygote mated with wild-type controls were identified as the deletion type using TA cloning (TransGen, China). Homozygous individuals were obtained by crossing male and female individuals identified with the same deletion type and assaying them further by TA cloning to confirm homozygosity. RNA Isolation and Quantitative Real-time PCR Total RNA from different developmental stages and various tissues was extracted using Trizol reagent (Thermo Fisher, USA). Genomic DNA was removed from total RNA using RNase-free DNase I (Takara, Dalian, China). Total RNA concentration and purity were assessed using a Multiskan GO microplate spectrophotometer (Thermo Fisher, USA). cDNA was synthesized using 1 μg of total RNA with the Prime Script Kit (Promega, USA). Specific primers are listed in Supporting Information (supplementary table S7, Supplementary Material online). The coding genes of B. mori ribosomal protein 49 (GenBank accession no. XM_004928077) and P. xuthus ribosomal protein 32 (GenBank accession no. NM_001312851) served as internal controls. Relative differences in transcript levels were analyzed using the ΔΔCt method, with six biological replicates and three technical replicates performed for each sample. In Vitro Wing Disc Culture Wing discs from L5D3 were aseptically dissected and incubated in Grace’s insect medium with or without 20E (Sigma-Aldrich, USA). The in vitro culture of wing discs was performed at 25 °C in Grace’s medium (Thermo Fisher, USA) following established protocols (Wang et al. 2009). Antibody Generation and Western Blot Analysis Specific sequences from BmApoD2 were cloned into the pCold1 expression vector (TaKaRa, Japan). The primers are listed in Supporting Information (supplementary table S7, Supplementary Material online). Recombinant proteins were expressed in E.coli BL21 (DE3) competent cells induced by isopropyl β-D-1-thiogalactopyranoside at 37 °C for 6 h. Polyclonal antibodies were generated using purified proteins (HuaBio, China). Mouse anti-His (Qiagen, USA), anti-AMPK (ab80039, Abcam, UK), antitubulin (Wang et al. 2009), horseradish peroxidase–conjugated antimouse and antirabbit IgG goat antibody (Southern Biotech, UAB), and anti-phosphorylated AMPK (#2535, Cell Signaling Technology, USA) were used in the experiments. Samples were homogenized in RIPA lysis buffer (Roche, USA), and total protein concentrations were determined using the BCA Protein Assay Kit (Takara, Japan). Protein separation was conducted on sodium dodecyl–sulfate polyacrylamide gel, followed by transfer onto a polyvinylidene difluoride membrane (0.45 μm; Millipore, USA). After blocking with 5% milk in Tris-buffered saline with 0.1% Tween 20 (TBST) for 2 h at 26 °C, the membrane was incubated with primary antibodies in 5% milk in TBST overnight at 4 °C. Following three washes with TBST for 10 min each, the membrane was incubated for 1 h at 26 °C with secondary antibodies in 5% milk in TBST. Immunoreactive signals were detected using the FDbio-Dura ECL Kit (FD-bio Science, China) and a chemiluminescence imaging system (Clinx, China). Expression of Recombinant ApoD Proteins Spodoptera frugiperda 9 (Sf 9) cells were cultured at 27 °C in Sf-900 (Thermo Fisher, USA) without fetal bovine serum. Recombinant BmApoD2, BmApoD3, and Trunc-BmApoD2 (with deletion of the longer segment, 191 to 285 amino acids, in the C terminus) proteins were expressed using the Bac-to-Bac system (Thermo Fisher, USA). Bacmid DNAs of recombinant baculoviruses were transfected into Sf-9 cell monolayers using an established protocol (Dai et al. 2021). Nickel chromatography (GE Healthcare, USA) was used to purify BmApoD recombinant proteins under denaturing conditions. Western blot analysis, using an anti-His antibody, was employed to confirm the specificity of purified recombinant proteins. Protein concentrations were determined using the Coomassie Plus Protein Assay Kit (Pierce, USA). Ligand-binding Assays Fluorescence measurements were conducted in a Shimadzu RF-5301PC spectrofluorometer (Ruiz, Sanchez et al. 2013; Zhou et al. 2018). Recombinant BmApoD2, BmApoD3, and Trunc-BmApoD2 were diluted to 1 μM in 10 mM phosphate buffer, 150 mM NaCl, and 1 mM ethylenediaminetetraacetic acid at pH 7.0. RA, cholesterol, ergosterol, and 20E were dissolved in N,N-dimethylformamide (Sigma-Aldrich, USA) to produce concentrated stock solutions within their solubility range. The mixture was equilibrated for 1 to 2 min in the dark after adding each ligand aliquot, and fluorescence was measured using an equilibrated mixture. Fluorescence intensity was measured at an excitation wavelength of 295 nm (selective for tryptophan residues), with the emission wavelength at 327 to 400 nm. The peak at 340 nm was used to calculate the apparent equilibrium dissociation constant. Data analysis was performed using the RFPC software package (Shimadzu, Japan). SPR Recombinant BmApoD1, BmApoD2, BmApoD3, or Trunc-BmApoD2 protein was immobilized on the sensor chip (CM5) using the amine-coupling method according to standard protocols. SPR was determined using a Biacore 8 K instrument (Cytiva, USA). The kinetics and affinity assay were examined at 25 °C at a flow rate of 30 µL/min using phosphate-buffered saline (PBS) buffer with 0.05% Tween-20. The data were collected using multicycle kinetics or single-cycle kinetics with the kinetics analysis (1:1 model) of Biacore 8 K evaluation software. Immunostaining and Tissue Staining Wings at P3 were dissected, fixed with 4% polyformaldehyde in PBS buffer (10 mM NaH2PO4/Na2HPO4 and 175 mM NaCl at pH 7.4) at 4 °C for 24 h, and dehydrated in 30% sucrose solution for 48 h at 4 °C. Wing disc tissues were embedded in tissue freezing medium (Thermo Fisher, USA) at <−20 °C to obtain frozen sections. Frozen slices were blocked in PBS with 3% bovine serum albumin for 1 h and incubated with anti-BmApoD2 antibody (1:200) overnight at 4 °C. After staining with fluorescein isothiocyanate–labeled goat antirabbit immunoglobulin secondary antibody (Thermo Fisher, USA) (1:200), frozen slices were counterstained with 4′,6-diamidino-2-phenylindole (DAPI; C1002, Beyotime, China) for 3 min. Images were captured using a laser scanning confocal microscope (ZEISS LSM780, Germany). Quantification was achieved through measurements taken in ImageJ (v1.53). P3 wings were dissected for frozen sections. Frozen slices were rinsed twice with PBS, stained with Nile red (Sangon Biotech, China) at a final concentration of 0.5 μg/mL for 30 min, and counterstained with DAPI for 3 min. Wing Area Measurement Wings from three groups, each containing 100 adult moths (50 males and 50 females), were dissected for wing area quantification. Data were analyzed using the “analyze particles” program in ImageJ (v1.53). Transcriptomic Analysis Three biological replicates were performed for each treatment. Sequencing was performed by BGI using a BGISEQ-500 sequencer (BGI, Shenzhen, China). Raw data were trimmed using SOAP nuke and Trimmomatic (Cock et al. 2010). Trimmed reads were mapped to the B. mori reference genome (GCF_014905235) using HISAT2 (Kim et al. 2019) or STAR (Dobin et al. 2013). Principal component analysis (PCA) was conducted using the princomp function in R (v4.1.1). DESeq2 packages (Love et al. 2014) were used for statistical significance determination based on read counts. A false-discovery rate (FDR) of < 0.001 and a |log2 (fold-change)| of >1.0 indicated significant differences. Unique id and id transfer were calculated using R scripts (v4.1.1) and the bitr function in clusterProfiler (Yu et al. 2012). For all GSEAs in nonmodel species, such as B. mori, annotation was acquired using the AnnotationHub packages (v4.1). Enrichment analysis was performed using clusterProfiler (Yu et al. 2012). Metabolomic Analysis Wings of BmApoD2−/− mutants and the wild type at P3 were collected and frozen in liquid nitrogen. Reversed-phase separation was performed using an ACQUITY UPLC BEH C18 column (Waters, UK) at 50 °C. Metabolites eluted from the column were detected using a high-resolution tandem mass spectrometer (Xevo G2 XS QTOF; Waters, UK). Extracted multiple reaction monitoring peaks were integrated using Progenesis v2.2 (Waters, UK). Metabolites with a relative SD of >30% were selected for subsequent analysis. Between-group differences in the abundance of liquid chromatography–mass spectrometry (LC–MS) identified metabolites were tested using metaX (Wen et al. 2017). Unsupervised PCA was employed to explore data quality, and supervised PLS-DA was applied to determine the number of responsible metabolites. Metabolites meeting the requirements of VIP ≥ 1, fold-change ≥ 1.2 or ≤ 0.8333, and q < 0.05 were defined as distinct samples. Functional analyses were based on the KEGG database. Metabolites were visualized using ComplexHeatmap (Gu et al. 2016). Proteomic Analysis Wings from BmApoD2−/− and the wild type at P3 were collected and frozen in liquid nitrogen. Following a bicinchoninic acid assay, normalized protein quantities were reduced using 10 mM dithiothreitol at 56 °C for 1 h, alkylated using 55 mM iodoacetamide in the dark at room temperature for 45 min, and digested overnight at a protein/trypsin ratio of 40:1. The resulting peptides were labeled with the iTRAQ 8plex Kit according to the manufacturer’s instructions. Labeled peptides were loaded on a trap column at 5 μL/min for 8 min and subsequently eluted into a homemade nanocapillary C18 column. The buffer B (98% acetonitrile and 0.1% formic acid) gradient was increased from 5% to 25% in 40 min, subsequently increased to 35% in 5 min, followed by a 2 min linear gradient increase to 80% and maintenance at 80% B for 2 min, and finally reduced to 5% in 1 min with equilibration for 6 min. The peptides separated using nanoHPLC were subjected to tandem MS (Q EXACTIVE HF X; Thermo Fisher Scientific, USA) for data-dependent acquisition detection via nanoelectrospray ionization. The raw files were analyzed using Proteome Discoverer and Mascot 2.3.02. For database searching, the following parameters were used: peptide mass tolerance = 20 ppm, fragment mass tolerance = 0.05 Da, carbamidomethyl (C), iTRAQ 8plex (N-term), and iTRAQ 8plex (K) of fixed modifications, I-KzbRa004 (19,893 sequences) database. Quantitative data were extracted from labeled peptides with isobaric tags using IQuant (Wen et al. 2014). Proteins exhibiting a 1.2-fold-change and P < 0.05 were defined as DEPs. ETC Complex I Activity Assay and ATP Level Determination Wings were homogenized using extraction buffer from the ETC Complex I Activity Assay Kit (BC0515, Solarbio, China) according to the manufacturer’s instructions. Complex I activity was calculated by detecting the NADH oxidation rate at 340 nm. ATP levels were determined using the ATP Activity Assay Kit (BC0300, Solarbio, China). NADPH, with an absorption peak at 340 nm, is considered directly proportional to ATP content; therefore, the NADPH level is considered to reflect ATP content. Lipidomics Analysis Briefly, 25 mg samples of wings at P3 were weighed and transferred into 2-mL tubes. Two small steel balls were added, as well as 800 μL of prechilled dichloromethane/methanol (3:1, v/v) and 10 μL of internal standard solution. The mixture was ground in a tissue grinder for 5 min, followed by ultrasonic treatment at 4 °C for 10 min. Subsequently, it was stored at −20 °C overnight, followed by centrifugation at 25,000 g for 15 min. After centrifugation, 600 μL of supernatant was removed from the vial for lyophilization and reconstituted using 600 μL of reconstitution solvent (isopropanol/acetonitrile/water = 2:1:1). The solution was vortexed for 10 min, followed by ultrasonic treatment for an additional 10 min, and then centrifuged for 15 min at 4 °C and 25,000 g. The pellet was discarded, and the supernatant was transferred to a 1.5-mL vial. To assess the repeatability and stability of LC–MS analysis, 10 μL was blended into a quality control sample. The lipids in the samples were determined using ultra-performance LC with tandem MS. Data processing was performed using LipidSearch (version 4.1; Thermo Scientific, USA). Differential lipid molecules were selected according to the criteria of fold-change ≥ 1.2 or ≤ 0.83 and P < 0.05. Additional Information Fourteen extended data figures and 7 supplementary tables for this paper are available online. Supplementary Material