Metabolite Changes in Orange Dead Leaf Butterfly Kallima inachus during Ontogeny and Diapause

Holometabolism is a form of insect development which includes four life stages: egg, larva, pupa, and imago (or adult). The developmental change of whole body in metabolite levels of holometabolous insects are usually ignored and lack study. Diapause is an alternative life-history strategy that can occur during the egg, larval, pupal, and adult stages in holometabolous insects. Kallima inachus (Lepidoptera: Nymphalidae) is a holometabolous and adult diapausing butterfly. This study was intended to analyze metabolic changes in K. inachus during ontogeny and diapause through a non-targeted UPLC-MS/MS (ultra-performance liquid chromatograph coupled with tandem mass spectrometry) based metabolomics analysis. A variety of glycerophospholipids (11), amino acid and its derivatives (16), and fatty acyls (nine) are crucial to the stage development of K. inachus. 2-Keto-6-acetamidocaproate, N-phenylacetylglycine, Cinnabarinic acid, 2-(Formylamino) benzoic acid, L-histidine, L-glutamate, and L-glutamine play a potentially important role in transition of successive stages (larva to pupa and pupa to adult). We observed adjustments associated with active metabolism, including an accumulation of glycerophospholipids and carbohydrates and a degradation of lipids, as well as amino acid and its derivatives shifts, suggesting significantly changed in energy utilization and management when entering into adult diapause. Alpha-linolenic acid metabolism and ferroptosis were first found to be associated with diapause in adults through pathway analyses. Our study lays the foundation for a systematic study of the developmental mechanism of holometabolous insects and metabolic basis of adult diapause in butterflies.


Introduction
Holometabolism is a form of insect development. It means insects will go through several developmental stages (immature and mature stages) to adapted for different activities during ontogeny [1,2]. The majority of described insects are holometabolous [3]. Compare with paurometabolous insects, immature stages (i.e., larvae and pupae) of holometabolous insects are very different from the mature stages (i.e., adults and diapausing individuals) on morphology, behavior, biology, and biochemistry [4]. Metabolomics is an important analytical tool of systems biology, which can identify all detectable metabolites in a biological system [5]. It is used to reveal endogenous metabolite changes that are caused by environment stress, disease process, or gene function [6]. Several metabolomes have been generated to study the developmental changes for the hemolymph of lepidopteron. In tobacco hornworm (Manduca sexta), 1 H-NMR result showed that the levels of many small molecule metabolites such as alanine, succinate, and fatty acid dramatically change in hemolymph during ontogeny [7]. Several studies suggest that the silkworm (Bombyx mori) exhibits specific metabolic pattern during ontogeny [8,9]. Lack of whole-body untargeted metabolome has existed as a problem for many years due to the limitations of 1 H-NMR,

Insect Collection
About 200 disease-free eggs of orange dead leaf butterfly (K. inachus) were acquired from the butterfly breeding base on Leshan. The egg, larva, pupal and adult stages of experimental butterflies were reared under the natural condition (12 h/12 h light/dark cycle;~20-25 • C; 70% relative humidity) in a net house (8 m long, 5 m wide, and 3 m tall) with evenly scattered sunlight. Larvae of K. inachus were reared on fresh leaves of Strobilanthes cusia, and the adults were fed 15% honey water. We used male and female larvae together since they were not readily distinguishable. K. inachus has the typical four-stage insect life cycle including egg, larval, pupal, and adult stages. After the eggs hatch, the larvae feed on the host plant, and when fully developed, pupate in a chrysalis. When metamorphosis is complete, the pupal skin splits, the adult insect climbs out. The molting time (four times) and the surficial red spines were used to identify fifth instar larvae used in this study. In this case, female adults initiate oocyte development within 4-5 days and have mature oocytes at 10-11 days after eclosion. Thus, inactive females with undeveloped oocytes on 10-11 day posteclosion were considered as individuals entering the fifth day of diapause. Samples were collected on first day of the fifth instar (LA group), fifth day after pupation (PU group), tenth day after eclosion (males, KM group; females, KF group), and fifth day after entering of diapause (female, dKF group), and six biological replicates were established. See Table 1 for the specific information of the samples. The samples were snap frozen simultaneously to limit non-biological sources of variation. Snap frozen samples were immediately stored at −80 • C.

Metabolite Extraction for the Untargeted Metabolomic Analysis
The sample stored at −80 • C refrigerator was thawed on ice. The thawed sample was homogenized by a grinder (30 HZ) for 20 s. A 400 µL solution (Methanol: Water = 7:3, v/v) containing internal standard was added in to 20 mg grinded sample, and shaked at 1500 rpm for 5 min. After placing on ice for 15 min, the sample was centrifuged at 12,000 rpm for 10 min (4 • C). A 300 µL of supernatant was collected and placed in −20 • C for 30 min. The sample was then centrifuged at 12,000 rpm for 3 min (4 • C). A 200 µL aliquots of supernatant were transferred for LC-MS analysis.
LIT and triple quadrupole (QQQ) scans were acquired on a triple quadrupole-linear ion trap mass spectrometer (QTRAP), QTRAP ® LC-MS/MS System (Applied Biosystems 4500 Q TRAP, San Diego, CA, US), equipped with an ESI Turbo Ion-Spray interface, operating in positive and negative ion mode and controlled by Analyst 1.6.3 software (AB Sciex, Framingham, US). The ESI source operation parameters were as follows: source temperature 500 • C; ion spray voltage (IS) 5500 V (positive), −4500 V (negative); ion source gas I (GSI), gas II (GSII), and curtain gas (CUR) were set at 55, 60, and 25.0 psi, respectively; the collision gas (CAD) was high. Instrument tuning and mass calibration were performed with 10 and 100 µmol/L polypropylene glycol solutions in QQQ and LIT modes, respectively. A specific set of MRM transitions were monitored for each period according to the metabolites eluted within this period. The MS RAW files for metabolite analysis have been deposited to the EMBL-EBI MetaboLights database with the identifier MTBLS5736, and can be accessed with the URL (http://www.ebi.ac.uk/metabolights/MTBLS5736 (accessed on 8 August 2022)).

Bioinformatic Analysis of the Untargeted Metabolomic Dataset
Unsupervised PCA (principal component analysis) was performed by statistics function prcomp within R (version 3.5.1, www.r-project.org (accessed on 25 March 2021)). The data was unit variance scaled before unsupervised PCA. In this process, dKF6 was deleted from the dKF group due to sample degradation and did not participate in the subsequent analysis. Significantly regulated metabolites between groups were determined by VIP ≥ 1 and absolute Log 2 FC (fold change) ≥ 1. VIP values were extracted from OPLS-DA result, which also contain score plots and permutation plots, was generated using R package MetaboAnalystR version 1.0.1 (McGill, QC, Canada). The data was log transform (log2) and mean centering before OPLS-DA. In order to avoid overfitting, a permutation test (200 permutations) was performed. Identified metabolites were annotated using KEGG compound database (http://www.kegg.jp/kegg/compound/ (accessed on 27 March 2021)), annotated metabolites were then mapped to KEGG pathway database (http://www.kegg.jp/kegg/pathway.html (accessed on 28 March 2021)). Significantly enriched pathways are identified with a hypergeometric test's p-value for a given list of metabolites. The figure of top 10 pathways was draw using the Sangerbox tools, a free online platform for data analysis (http://vip.sangerbox.com/ (accessed on 28 March 2021)). The reliability of markers was calculated by calculating formulas as follows: R = [SD (KF)/mean (KF) + SD (dKF)/mean (dKF)].

Statistical Analysis
Significance of differences between groups was tested using unpaired t-tests. Correlations between variables were calculated using Pearson's product-moment correlation. Both tests were performed using GraphPad Prism version 8.07 (GraphPad Software, La Jolla, CA, USA).

Untargeted Metabolite Analysis of Samples
To explore global metabolic variations, an untargeted metabolomic approach was used (n = 29, Table 1), which identified 1683 annotated metabolites from 3587 ion features (Table S1). Based on their annotations, plentiful metabolites were assigned to at least one primary or secondary metabolic category. The top 20 largest secondary metabolic categories, such as small peptide (191 metabolites), amino acid derivatives (85 metabolites), acyl carnitines (64 metabolites), nucleotide and its derivatives (54 metabolites), lysophosphatidyl choline (35 metabolites), and organic acid and its derivatives (33 metabolites) are shown in Figure S1.
To provide a deep overview of the metabolic variations, several quality control parameters for the quantification, including coefficient of variation (CV), principal component (PC) and orthogonal partial least square-discriminate analysis (OPLS-DA) were analyzed [34]. PCA score plot of these features have shown that a clear separation among different groups, with 69.8% of the variation explained by the first two principal component axes ( Figure 1A, PC1 and PC2 explained 42.9% and 26.9%, respectively). To examine the quality of the acquired QC data, CV (coefficient of variation) analysis were generated for all of the samples and revealed a high degree of overlap using empirical cumulative distribution function (ECDF). From the Figure 1B we can see that the proportion of metabolites with CV value less than 0.5 in all QC samples was higher than 85%. The OPLS-DA was performed to identify the metabolites responsible for the separation between different successive stages. The score plot of OPLS-DA shows a clear separation in four comparison groups (LA vs. PU, PU vs. KM, PU vs. KF, KM vs. KF). To statistically validate the model, the permutation test (200 permutations) was performed. For LA vs. PU, R 2 X, R 2 Y, and Q 2 were 0.685, 1.000, and 0.999, respectively; For PU vs. KM, R 2 X, R 2 Y, and Q 2 were 0.893, 1.000, and 0.999, respectively; For PU vs. KF, R 2 X, R 2 Y, and Q 2 were 0.883, 1.000, and 0.999, respectively; For KM vs. KF, R 2 X, R 2 Y, and Q 2 were 0.447, 1.000, and 0.477, respectively ( Figure 2A different groups, with 69.8% of the variation explained by the first two principal component axes ( Figure 1A, PC1 and PC2 explained 42.9% and 26.9%, respectively). To examine the quality of the acquired QC data, CV (coefficient of variation) analysis were generated for all of the samples and revealed a high degree of overlap using empirical cumulative distribution function (ECDF). From the Figure 1B we can see that the proportion of metabolites with CV value less than 0.5 in all QC samples was higher than 85%. The OPLS-DA was performed to identify the metabolites responsible for the separation between different successive stages. The score plot of OPLS-DA shows a clear separation in four comparison groups (LA vs. PU, PU vs. KM, PU vs. KF, KM vs. KF). To statistically validate the model, the permutation test (200 permutations) was performed. For LA vs PU, R 2 X, R 2 Y, and Q 2 were 0.685, 1.000, and 0.999, respectively; For PU vs. KM, R 2 X, R 2 Y, and Q 2 were 0.893, 1.000, and 0.999, respectively; For PU vs. KF, R 2 X, R 2 Y, and Q 2 were 0.883, 1.000, and 0.999, respectively; For KM vs. KF, R 2 X, R 2 Y, and Q 2 were 0.447, 1.000, and 0.477, respectively ( Figure 2A). Based on the results of OPLS-DA, 1599 high-quality metabolites were used to screen the differentially accumulated metabolites (DAMs), which were determined by variable importance in projection (VIP) ≥ 1 and absolute Log2FC (fold change)

Vital Metabolites for Stage Development
All of the DAMs were assigned to various primary metabolic categories, including lipids, fatty acyl, heterocylic compounds, organicacid and its derivatives, oxidized lipids, carboxylic acids and derivatives, nucleotide and its metabolomics and others. The global metabolic captured the 120 DAMs with significant difference in two transitions of successive stages (larva to pupa and pupa to adult).
As shown in Figure 3A, plenty of glycerophospholipids 11 metabolites such as lysope 18:0, pysoPE 20:2, and pysoPE 18:2 were predominantly accumulated in larval stage, while several carboxylic acids or its derivatives (three metabolites), nucleotide or its metabolomics (three metabolites), and benzene and substituted derivatives (five metabolites) have higher levels in larva than other stages. For pupal stage, the most striking characterization is abundant amino acid and its derivatives (16 metabolites, such as L-tryptophan and L-glutamate) which did not maintain in the former or the latter stages, while the levels of four heterocyclic compounds peak in this stage. For ault stage, in addition to the sharp drop in the levels of various amino acid or its metabolomics after pupal stage, the most notable was the large accumulation of multiple fatty acyls (nine metabolites). Furthermore, there were no significant differences in vital metabolic categories for transition of pupal to adult stages between males and females.

Vital Metabolites for Stage Development
All of the DAMs were assigned to various primary metabolic categories, including lipids, fatty acyl, heterocylic compounds, organicacid and its derivatives, oxidized lipids, carboxylic acids and derivatives, nucleotide and its metabolomics and others. The global metabolic captured the 120 DAMs with significant difference in two transitions of successive stages (larva to pupa and pupa to adult).
As shown in Figure 3A, plenty of glycerophospholipids 11 metabolites such as lysope 18:0, pysoPE 20:2, and pysoPE 18:2 were predominantly accumulated in larval stage, while several carboxylic acids or its derivatives (three metabolites), nucleotide or its metabolomics (three metabolites), and benzene and substituted derivatives (five metabolites) have higher levels in larva than other stages. For pupal stage, the most striking characterization is abundant amino acid and its derivatives (16 metabolites, such as L-tryptophan and Lglutamate) which did not maintain in the former or the latter stages, while the levels of four heterocyclic compounds peak in this stage. For ault stage, in addition to the sharp drop in the levels of various amino acid or its metabolomics after pupal stage, the most notable was the large accumulation of multiple fatty acyls (nine metabolites). Furthermore, there were no significant differences in vital metabolic categories for transition of pupal to adult stages between males and females.   Table 2 shows the top five DAMs (up-or down) identified based on log2FC in transition of successive stages and their related metabolic pathways. In LA vs. PU, 2-Keto-6acetamidocaproate, carnitine C13:1, N-phenylacetylglycine, 3-Chloro-L-tyrosine, and carnitine C14:1-OH are the DAMs with the maximum Log2FC value; Cinnabarinic acid, trans, cis-3,6-nonadien-1-ol, 2-(formylamino) benzoic acid, 13,14-Dihydro-15-keto-PGD2, and (Z)-6-octadecenoic acid are the DAMs with the minimum Log2FC value. In PU vs. KF (stands for imagoes), DL-O-tyrosine, tauropine, L-palmitoylcarnitine, carnitine C16:0, and L-histidine are the DAMs with the maximum Log2FC value; L-glutamate, L-glutamine, ®-(-)-2-phenylglycine, L-isoisoleucine, and L-tryptophan are the DAMs with the minimum Log2FC value. In KM vs. KF, 4-O-Dimethylallyl-L-tyrosine, N-lactoyl-phenylalanine, Salicin-6P, Uridine triphosphate, and 3-Hydroxypicolinic acid are the DAMs with the maximum Log2FC value; 5'-deoxyadenosine, 2'-deoxyadenosine, S-Sulfo-L-cysteine, protocatechuic aldehyde, carnitine C9:2-OH are the DAMs with the minimum Log2FC value.

The Endogenous Metabolome of Different Developmental Stages
The endogenous metabolome is the set of endogenous metabolites in biological samples, which can reflect the endogenous mechanisms of adaptation and the small-molecule chemicals that play a role in survival within species [35]. In this study, the endogenous metabolome of larval, pupal, and adult (males and females) stages in K. inachus were generated by LC-MS. The top five small-molecule chemicals (heavyweights) with the highest content at each stage were significantly different between immature and mature stages ( Figure 3B). After searching the metabolite pool of LA groups, N-acetylglycine, pysoPE 18:2, L-histidine, pysoPE 18:3, and (1R, 6S)-6-amino-5-oxocyclohexyl-2-ene-1-carboxylic acid ester (CAE) were identified as heavyweights in larval stage. When entering the pupal stage, three heavyweights ((1R, 6S)-CAE, L-histidine, and N-acetylglycine) in larval stage were retained but their ranking changed. Two new metabolites (betaine and Glycerol-3-phosphate) were identified as heavyweights in this stage. In adult stage, the top four metabolites ((1R, 6S) -CAE, L-tyrosine, 6-aminocaproic acid, and L-valine) in both males and females are identical, while the fifth heavyweight are 6-Methyl mercaptopurine in females and L-isoleucine in males. The bar graph above shows the major characteristic of the endogenous metabolome is the high assay of (1R, 6S) -CAE during ontogeny in K. inachus ( Figure 3C). The content of L-tyrosine was the second in adult stage. Furthermore, the average molecular weight of heavyweights gradually decreases as life cycle progresses (284.74, 149.25, 144.06, and 143.08 Da in larvae, pupae, females, and males, Table S2).

Diapause-Associated Changes in Metabolite Profiles
The aforementioned analyses provide overall changes in metabolite profiles during ontogeny in K. inachus. The next part of this study was concerned on the biomarkers of entering the diapause maintenance and main metabolic pathways enriched by DAMs identified between KF and dKF groups. The metabolic pools of KF and dKF were compared in order to study the difference between the metabolome in diapause and non-diapause states of K. inachus adult female. The score plot of OPLS-DA shows a clear separation of diapause and non-diapause samples ( Figure 4A). To statistically validate the model, the permutation test (200 permutations) was performed. The R 2 X and R 2 Y of the OPLS-DA model was 0.87 and 1.00, indicating the explanatory rate of the model to the X and Y matrix is high. The Q 2 was 0.999, indicating a good predictivity. Our untargeted metabolomics dataset ( Figure 4B) included 614 metabolites that were less abundant (log 2 FC ≤ −1, p ≤ 0.05) and 520 metabolites that were more abundant (log 2 FC ≥ 1, p ≤ 0.05) in diapause adult females relative to non-diapause adult females. The top 20 heavyweights with the highest content at diapause individuals were highly taxonomically concentrated. Among them, 50% of the heavyweights were glycerophospholipids and 45% were amino acids or AADs ( Figure 4C). Further analysis showed that more than 90% of DAMs could be classified into 12 primary categories in main metabolite profiles ( Figure 4D).

Diapause-associated Changes in Metabolite Profiles
The aforementioned analyses provide overall changes in metabolite profiles during ontogeny in K. inachus. The next part of this study was concerned on the biomarkers of entering the diapause maintenance and main metabolic pathways enriched by DAMs identified between KF and dKF groups. The metabolic pools of KF and dKF were compared in order to study the difference between the metabolome in diapause and non-diapause states of K. inachus adult female. The score plot of OPLS-DA shows a clear separation of diapause and non-diapause samples ( Figure 4A). To statistically validate the model, the permutation test (200 permutations) was performed. The R 2 X and R 2 Y of the OPLS-DA model was 0.87 and 1.00, indicating the explanatory rate of the model to the X and Y matrix is high. The Q 2 was 0.999, indicating a good predictivity. Our untargeted metabolomics dataset ( Figure 4B) included 614 metabolites that were less abundant (log2FC ≤ −1, p ≤ 0.05) and 520 metabolites that were more abundant (log2FC ≥ 1, p ≤ 0.05) in diapause adult females relative to non-diapause adult females. The top 20 heavyweights with the highest content at diapause individuals were highly taxonomically concentrated. Among them, 50% of the heavyweights were glycerophospholipids and 45% were amino acids or AADs ( Figure 4C). Further analysis showed that more than 90% of DAMs could be classified into 12 primary categories in main metabolite profiles ( Figure  4D).  Several DAMs were identified as potential biomarkers of entering the diapause maintenance based on the ®io (R) of standard deviation to mean in two groups. Several representative biomarkers belong to glycerophospholipids and amino acids or AADs are presented in Table 3.

Key Pathways Affected by Entering into Adult Diapause
Through the KEGG pathway analysis of the DAMs, the variational metabolic pathways caused by entering diapause maintenance in K. inachus adult females were analyzed. The enrichment of pathways is presented in Figure 5. Over 300 metabolites across more than 20 pathways, including the metabolism of lipids, amino acids, carbohydrates and coenzymes were changed in diapause individuals. Detailed enriched pathway information is provided in Table S3. L-glutamic acid (MEDN0011), L-phenylalanine (MEDN0015), L-tyrosine (MEDP0009), L-glycine (MEDP0006), and L-cysteine (MEDP0052), etc., are the metabolites enriched to multi-pathways, suggesting that the changes of major metabolites during diapause in K. inachus are complex and multi-pathway cooperative. Detailed metabolite information is provided in Table S1.

Discussion
The mechanism of complete metamorphosis is one of the most important challenges in the study of holometabolism. However, the metabolite changes during stage transitions have not been systematically studied in holometabolous insects. Similarly, data on the accumulation and degradation of metabolites during adult diapause are scarce, and the lack of biomarkers make it difficult to identify diapause adults occurring in natural state. Therefore, in this study we identified vital metabolites for stage development and transition of successive stages in K. inachus. Among the four comparable groups (LA vs. PU, PU vs. KF, PU vs. KM, and KF vs. KM), there were more DAMs in the PU vs. KM (782) or KF (840) than in the LA vs. PU comparisons (362), indicating that eclosion of K. inachus pupae results in expression changes of many DAMs although pupation dramatically changes the insect's appearance physiology and functional structure [36]. The reason for this phenomenon is likely to be the intensity of the endocrine system affected by microRNA varies between pupation (high) and eclosion (low) [37].
For larval stage development, 11 glycerophospholipids (seven lysophosphatidylcholines and three lysophosphatidylethanolamines) were found to accumulate specifically in the larval stage of K. inachus. Moderate accumulation of glycerophospholipids has also been observed in first-instar of other holometabolic insects [38]. Glycerophospholipids are the main component of biological membranes. The glycerophospholipids metabolism was

Discussion
The mechanism of complete metamorphosis is one of the most important challenges in the study of holometabolism. However, the metabolite changes during stage transitions have not been systematically studied in holometabolous insects. Similarly, data on the accumulation and degradation of metabolites during adult diapause are scarce, and the lack of biomarkers make it difficult to identify diapause adults occurring in natural state. Therefore, in this study we identified vital metabolites for stage development and transition of successive stages in K. inachus. Among the four comparable groups (LA vs. PU, PU vs. KF, PU vs. KM, and KF vs. KM), there were more DAMs in the PU vs. KM (782) or KF (840) than in the LA vs. PU comparisons (362), indicating that eclosion of K. inachus pupae results in expression changes of many DAMs although pupation dramatically changes the insect's appearance physiology and functional structure [36]. The reason for this phenomenon is likely to be the intensity of the endocrine system affected by microRNA varies between pupation (high) and eclosion (low) [37].
For larval stage development, 11 glycerophospholipids (seven lysophosphatidylcholines and three lysophosphatidylethanolamines) were found to accumulate specifically in the larval stage of K. inachus. Moderate accumulation of glycerophospholipids has also been observed in first-instar of other holometabolic insects [38]. Glycerophospholipids are the main component of biological membranes. The glycerophospholipids metabolism was found to be related to the lipid regulation, lipoprotein, whole-body energy metabolism and metabolic disorders [39]. Cellular lipids have been reported to play crucial roles in the replication of a number of viruses, including enveloped viruses and no-enveloped viruses [40]. This may be the metabolic basis of the larval stage as the susceptibility stage of nuclear polyhedrosis virus (NPV) in Lepidoptera. For pupal stage development, a variety of common amino acids specifically accumulate at this stage. Among them, L-tryptophan and L-glutamate may be more important for the chrysalis. L-tryptophan is a heterocyclic amino acid and the major limiting amino acids in most insects [41]. In the majority of holometabolous insects, most larval tissues and organs are actually re-specified, including the epidermis, nervous system and muscles [3]. It has been proved that the determinant of pupal viability of D. melanogaster cinnabar and cardinal eye color mutants is tryptophan metabolism [42]. Therefore, high levels of L-tryptophan may be associated with characteristic formation and eclosion during pupal stage of K. inachus. Notably, after about 20 days, tryptophan levels dropped to almost undetectable levels in imagoes. An acceptable assumption is that tryptophan is converted to nicotinic acid for preservation in adults, which is supported by metabolome data showing a 6-fold increase (p < 0.05) in nicotinic acid from pupal to adult stages (Table S1). In addition, prior studies that have noted the tryptophan was used to synthesize pigment in butterflies such as the papiliochromes [43]. This reference may support the hypothesis that tryptophan as precursors of 'supposed' pigment, which is indispensable to the formation of wing patterns of K. inachus. One unanticipated finding was the accumulation of L-glutamate in pupal stage because studies in fruit fly have found that L-glutamate is inactive in the pupal and larval stages [44]. L-glutamate mediates fast inhibitory neurotransmission by affecting glutamate-gated chloride channels in invertebrates [45]. Hence, the reason of immobile and slow-acting in chrysalis of K. inachus maybe fast inhibitory neurotransmission caused by the high content of L-glutamate. For adult stage development, levels of several free fatty acids were found to increase dramatically. Free fatty acids are the product of lipolysis of storage fat, i.e., triacylglyceroles [46]. As mentioned in the study of S. argyrostoma, the free fatty acids profile varies considerably between each development stage [47].
Fold change analysis showed that the drastic changes in the concentrations of the following four DAMs may be the trigger for pupation. (1) 2-Keto-6-acetamidocaproate is the intermediate of L-lysine to L-alpha-aminoadipate and may have the potential detoxification abilities to protect the K. inachus during pupal stage [48]. (2) N-phenylacetylglycine was found to scavenge extracellular peroxide (pyruvate), inhibit ROS and activate cellular antioxidant defense, act as indicators of antioxidant mobilization against oxidative stress [49].
(3) Cinnabarinic acid is an endogenous metabolite of the kynurenine pathway that meets the structural requirements to interact with glutamate receptor, [50]. (4) 2-(Formylamino) benzoic acid is an upstream metabolite of cinnabarinic acid, and the decrease of both is to accumulate L-kynurenine which promoter the degradation of tryptophan, and biosynthesis eye chromes to protect nervous system in insects (Log 2 FC = 3.50, p < 0.05) [51]. Three DAMs (including increased L-histidine and decreased L-glutamate and L-glutamine) related to histidine metabolism were identified in the transition of pupal to adult stages. Recent study showed that Lepidopteran wing scales contain abundant histidine-rich cuticular proteins [52]. This is one of the reasons why biosynthesis of histidine was promoted in adults. The regulation of histidine metabolism is the key to eclosion. Several DAMs related to nucleotide and amino acid metabolism have also been found between males and females. Therefore, we hypothesize that this phenomenon is due to female oogenesis earlier than male spermatogenesis. The function of the most of remaining top DAMs in insects remains unclear. However, it is necessary to study these metabolites in further study because of the great change of their levels during the stage transitions.
(1R, 6S) -CAE is a metabolin in phenazine biosynthes and an intermediate from chorismate to phenazine [53]. In this study, (1R, 6S) -CAE is the 1st metabolite in the adult and pupal stages but the fifth in the larval stage in content. Actually, the content of (1R, 6S) -CAE in larvae is 3-5 times (p < 0.05) that in other stages. However, the reason for the high level of (1R, 6S) -CAE in all developmental stages, especially larval stage, and its evolutionary background are still unclear. Large accumulations of this metabolite are not seen in other holometabolous insects until now. In contrast to (1R, 6S) -CAE, further statistical test revealed that L-tyrosine in adult is 3-5 times (p < 0.05) that in other stages. L-tyrosine is the substrates for melanin biosynthesis, which have been demonstrated to regulate color and morphology of butterfly wing scales [54]. The other two metabolites with high concentrations are N-acetylglycine and betaine, which were significantly different (p < 0.05) between larval and pupal stages. However, this difference is much smaller than the decline after entering the mature stage.
Furthermore, we also plotted the major metabolic profiles during diapause in K. inachus adult females and provided several potential biomarkers of diapause maintenance. Surprisingly, all DAMs belong to lipids were decreased in the dKF group, but most DAMs belong to GPs were detected to be highly up-regulated. This finding is consistent with that of R. cerasi diapause pupae which catabolize a variety of lipids during the summer diapause [19]. Meanwhile, female adults accomplished seasonal acquisition of chill tolerance by synthesizing large amounts of glycerophospholipids that are different from those accumulated during larval stage. This result is in line with those of previous study on P. apterus [55]. Similar to the case of glycerophospholipids, the level of most members of carbohydrates (carbs) DAMs was also found to been deeply raised. Increasing carbohydrate content is one of the strategies used by many insects to combat extreme weather conditions in both pupal and larval diapauses [56][57][58]. Combined with the results of this study, the increase of carbohydrate content is an important marker of early diapause maintenance in all stages in insects. Further, several reports have showed the importance of trehalose during diapause in invertebrates [59,60]. In present study, metabolome of dKF matched those observed in earlier studies of Yi et al. [33] who concluded the type of carbohydrate-alcohol accumulation of K.inachus was trehalose. For the category with the most detected DAMs, amino acids and AADs, the level changes of DAMs are slightly complicated but still most DAMs also rose steadily. Amino acids and their derivatives vary greatly during egg [61], larval [62] or adult [63] diapause. This discrepancy could be attributed to the metabolic system of amino acids to adapt to the needs of biosynthesis and environmental stress. In conclusion, DAMs belong to glycerophospholipids, carbs and amino acids and AADs in this study are suitable as detectable biomarkers for entering the diapause maintenance stage of adults.
KEGG enrichment analysis indicate several pathways are associated with adult diapause in K. inachus females. Regulating these targeted pathways may achieve purposeful intervention in the expression of metabolites. Most of the pathways enriched only accumulated one or two metabolites (converted from other metabolites of the pathway or from upstream pathways) in dKF groups. For example, in alpha-linolenic acid metabolism, the level of 9(S)-HOTrE increases with the decreasing level of alpha-linolenate, 13(S)-HpOTrE and others. 9(S)-HOTrE is an octadecanoid that could increase apolipoprotein transcription [64]. Active lipid metabolism could be observed in the metabolic profile above. A possible explanation for this might be that accumulation of 9(S)-HOTrE in diapause adults promotes the production of apolipoprotein and then regulates active lipid metabolism. In further study, we will investigate the differences in physical index (such as cold tolerance, lifespan, and activeness), transcriptome and methylome between the second and third generations of K. inachus which are lipoxygenase (a key enzyme of 9(S)-HOTrE) knockdown through transcriptome and whole genome bisulfite sequencing. Ferroptosis is a new type of cell death that was discovered in recent years and is usually accompanied by a large amount of iron accumulation and lipid peroxidation during the cell death process [65]. Larval-specific tissues (such as prothoracic glands and anterior silk glands) are normally eliminated by programmed cell death after pupation [66]. However, for imagoes, the accumulation of cysteine and arachidonate acid through Ferroptosis is likely to be the key to entering into diapause [67]. At present, there are few studies on alpha-linolenic acid metabolism and ferroptosis in diapause insects. The results can provide a reference for relevant studies of targeted pathways.

Conclusions
In this study, we investigated the metabolic changes in K. inachus during ontogeny and diapause by untargeted metabolomics (UPLC-MS/MS), analyzed the vital metabolites for stage development and transition of successive stages, and summarized potential biomarkers for entering into diapause stage of adults and key pathways affected by entering into adult diapause.
Glycerophospholipids, amino acid or its derivatives, and fatty acyls is crucial for larval, pupal, and adult stages of K. inachus, respectively. 2-keto-6-acetamidocaproate, N-phenylacetylglycine, Cinnabarinic acid, and 2-(Formylamino) benzoic acid are momentous in stage transition of larva to pupa; L-histidine, L-glutamate and L-glutamine are momentous in stage transition of pupa to imago; Levels of (1R, 6S) -CAE and L-tyrosine are high throughout development but the specific functions still need to be explored. Lipids degradation and synthesis of glycerophospholipids and carbohydrates were observed in diapause individuals, and the metabolism of amino acids and their derivatives was very active in diapause individuals. Alpha-linolenic acid metabolism and ferroptosis were first found to be associated with diapause in adults. Our study lays the foundation for a systematic study of the developmental mechanism of holometabolous insects and metabolic basis of adult diapause in butterflies.

Supplementary Materials:
The following supporting information can be downloaded at: https://www. mdpi.com/article/10.3390/metabo12090804/s1. Figure S1: The top 20 largest metabolic secondary categories of all identified metabolites. The Y-axis represents the number of metabolites. Table S1: Detailed information of the metabolites identified in untargeted metabolome, Table S2: The molecular weight of top five small-molecule chemicals with the highest content. Table S3. Detailed information of the KEGG pathway enriched by DAMs identified between KF and dKF groups.

Data Availability Statement:
The data presented in this study are available in article and supplementary material.

Conflicts of Interest:
The authors declare no conflict of interest.