Metabolic Changes in Larvae of Predator Chrysopa sinica Fed on Azadirachtin-Treated Plutella xylostella Larvae

Biological control is a key component of integrated pest management (IPM). To suppress pests in a certain threshold, chemical control is used in combination with biological and other control methods. An essential premise for using pesticides in IPM is to ascertain their compatibility with beneficial insects. Chrysopa sinica (Neuroptera: Chrysopidae) is an important predator of various pests and used for pest management. This study was intended to analyze metabolic changes in C. sinica larvae after feeding on azadirachtin-treated Plutella xylostella (Lepidoptera, Plutellidae) larvae through a non-targeted LC–MS (Liquid chromatography–mass spectrometry) based metabolomics analysis. Results showed that C. sinica larvae did not die after consuming P. xylostella larvae treated with azadirachtin. However, their pupation and eclosion were adversely affected, resulting in an impairment in the completion of their life cycle. Feeding C. sinica larvae with azadirachtin-treated P. xylostella larvae affected over 10,000 metabolites across more than 20 pathways, including the metabolism of amino acids, carbohydrates, lipid, cofactors, and vitamins in C. sinica larvae, of which changes in amnio acid metabolism were particularly pronounced. A working model was proposed to illustrate differential changes in 20 metabolites related to some amino acid metabolisms. Among them, 15 were markedly reduced and only five were elevated. Our results suggest that azadirachtin application may not be exclusively compatible with the use of the predator C. sinica for control of P. xylostella. It is recommended that the compatibility should be evaluated not only based on the survival of the predatory insects but also by the metabolic changes and the resultant detrimental effects on their development.


Introduction
Integrated pest management (IPM) is a coordinated process using multiple methods, such as biological, chemical, cultural, mechanical, physical, and pest resistant or tolerant varieties for optimizing control of pests in an ecologically and economically sound manner [1]. Among them, chemical control and biological control are two common methods used either separately or in combination for managing insect pests during crop production. However, the use of pesticides may have adverse effects on non-target organisms, including predatory insects [2]. In order to minimize these problems and maintain sustainable control of insect pests, botanical pesticides are considered to be attractive alternatives for pest management. Compared with traditional synthetic insecticides, botanical pesticides have better eco-toxicological properties, including low toxicity, rapid degradation, and little impact on the environment, which makes them a suitable choice for pest control [3,4].

Azadirachtin Activities against C. sinica
The larvae of C. sinica fed with P. xylostella larvae that consumed azadirachtin-treated leaves (T) showed no significant growth differences compared to those fed P. xylostella control (CK) as there were no larval mortalities between T and CK before pupation (data not shown). After stopping preying, mature larvae began to pupate. The head and tail of the larvae gradually curled together, and the tail drew silk to make cocoons (Figure 1(c-4)). However, there were 18% of the larvae in the T treatment that could not curl up and could not draw silk to make a cocoon (Figure 1(c-1-c-3)). When touching the larvae with a small brush, they still twisted but died in a few days. After 20 to 30 days, the C. sinica adults emerged from the pupae. In the T treatment, 24% of adults were deformed as they were unable to extend their wings and/or had malformed abdomen (Figure 1(d-1)), which were regarded as the failure to eclosion. Adults with fully extended wings and no growth defects were considered as successful eclosion (Figure 1(d-2,d-3)). Thus, the pupation and eclosion of C. sinica larvae were significantly affected by T treatment. As shown in Figure 2, the proportion of larvae undergoing pupation from T treatment was significantly lower at 82.00 ± 3.06% comparted to 100.00% in the CK (p < 0.01), and their eclosion was 76.00 ± 1.15% in the T treatment against 98.00 ± 1.48% of CK (p < 0.001).   The process of monitoring P. xylostella larvae feeding on cabbage leaves and C. sinica larvae ingesting azadirachtin-treated P. xylostella larvae and subsequently their pupation and eclosion. A larva feeding on a cabbage leaf treated with azadirachtin (a). A C. sinica larva ingesting a P. xylostella larva (b). C. sinica larvae underwent pupation from curling to the formation of cocoon (c-1-c-4). The emergence of C. sinica adult: a deformed adult (d-1), normal adult (d-2), and normal adult after removing wings to show normal abdomen (d-3).

Figure 2.
The proportion of mature C. sinica larvae underwent pupation (a) and eclosion (b) after ingesting azadirachtin-treated P. xylostella larvae. Data were expressed as the mean ± S.E. and ** and *** indicate significant differences at p < 0.01 and p < 0.001 levels based on Tukey's HSD test.

Metabolic Profiles Analyzed by LC-MS
The unsupervised PCA was used to check the quality of the data from the LC-MS analyses. In ESI+ mode, the PC1 and PC2 explained 30.9% and 11.4% of the total variance of all samples. In ESI− mode, the PC1 and PC2 explained 33.9% and 10.3% of the total variance. The supervised PLS-DA was performed to identify the metabolites responsible for the separation between CK and azadirachtin treatments (T). Results showed that in the ESI+ mode, the R 2 X, R 2 Y, Q 2 Y, and RMSEE values in the PLS-DA model were 0.59, 0.753, 0.369, and 0.269 (Figure 3a), respectively; in the ESI− mode, the R 2 X, R 2 Y, Q 2 Y, and RMSEE values in the PLS-DA model were 0.728, 0.74, 0.536, and 0.276 (Figure 3b), respectively. Based on the OPLS-DA model, the R 2 X, R 2 Y, Q 2 Y, and RMSEE values in the ESI+ mode were 0.59, 0.753, 0.337, and 0.269 (Figure 4a), respectively. In the ESI− mode, the R 2 X, R 2 Y, Q 2 Y, and RMSEE values in the ESI− mode were 0.728, 0.74, 0.512 and 0.276 (Figure 4b), respectively. ingesting azadirachtin-treated P. xylostella larvae. Data were expressed as the mean ± S.E. and ** and *** indicate significant differences at p < 0.01 and p < 0.001 levels based on Tukey's HSD test.

Metabolic Profiles Analyzed by LC-MS
The unsupervised PCA was used to check the quality of the data from the LC-MS analyses. In ESI+ mode, the PC1 and PC2 explained 30.9% and 11.4% of the total variance of all samples. In ESI− mode, the PC1 and PC2 explained 33.9% and 10.3% of the total variance. The supervised PLS-DA was performed to identify the metabolites responsible for the separation between CK and azadirachtin treatments (T). Results showed that in the ESI+ mode, the R 2 X, R 2 Y, Q 2 Y, and RMSEE values in the PLS-DA model were 0.59, 0.753, 0.369, and 0.269 (Figure 3a), respectively; in the ESI− mode, the R 2 X, R 2 Y, Q 2 Y, and RMSEE values in the PLS-DA model were 0.728, 0.74, 0.536, and 0.276 (Figure 3b), respectively. Based on the OPLS-DA model, the R 2 X, R 2 Y, Q 2 Y, and RMSEE values in the ESI+ mode were 0.59, 0.753, 0.337, and 0.269 (Figure 4a), respectively. In the ESI− mode, the R 2 X, R 2 Y, Q 2 Y, and RMSEE values in the ESI− mode were 0.728, 0.74, 0.512 and 0.276 (Figure 4b), respectively.

Changes in Metabolites and Metabolic Pathways of Differentially Abundant Metabolites
Representative LC-MS total ion chromatograms (TICs) of C. sinica larvae tissue samples are shown in Figure 5. The shape and quantity of peaks between the T and CK treatments varied greatly. Approximately 13,672 and 10,947 metabolite peaks were deconvoluted in ESI+ and ESI− modes of LC-MS, respectively. The ESI+ usually detects N, O, and S-containing species and also some specific hydrocarbons, such as isoprene, terpenes, and aromatics as protonated neutral MH+, whereas the ESI− detects acid including carboxylic acids RCOOH and inorganic acids and hydrosulfides as deprotonated neutral [M-H] − [42]. A total of 3210 and 2026 remaining peaks in ESI+ and ESI− modes in LC-MS were further annotated using references in existing databases, respectively. After the exogenous compounds in LC-MS were removed, the differentially abundant metabolites were selected according to the VIP values from the OPLS-DA model (VIP ≥ 1) and the corrected p values from t-test (p < 0.05). There were 778 compounds in the ESI+ model, of which 357 were

Changes in Metabolites and Metabolic Pathways of Differentially Abundant Metabolites
Representative LC-MS total ion chromatograms (TICs) of C. sinica larvae tissue samples are shown in Figure 5. The shape and quantity of peaks between the T and CK treatments varied greatly. Approximately 13,672 and 10,947 metabolite peaks were deconvoluted in ESI+ and ESI− modes of LC-MS, respectively. The ESI+ usually detects N, O, and S-containing species and also some specific hydrocarbons, such as isoprene, terpenes, and aromatics as protonated neutral MH+, whereas the ESI− detects acid including carboxylic acids RCOOH and inorganic acids and hydrosulfides as deprotonated neutral [M-H] − [42]. A total of 3210 and 2026 remaining peaks in ESI+ and ESI− modes in LC-MS were further annotated using references in existing databases, respectively. After the exogenous compounds in LC-MS were removed, the differentially abundant metabolites were selected according to the VIP values from the OPLS-DA model (VIP ≥ 1) and the corrected p values from t-test (p < 0.05). There were 778 compounds in the ESI+ model, of which 357 were upregulated and 421 were downregulated. In ESI− mode, there were 391 compounds: 180 were upregulated and 211 were downregulated.   Table 1 shows some related metabolic pathways with representative differentially abundant metabolites and their upregulation and downregulation. Amino acids, carbohydrates, bile, lipids, membrane transports, cofactors and vitamins were primary metabolites that were affected in C. sinica larvae after ingestion of azadirachtin-treated P. xylostella larvae. The enrichment of pathways is presented in Figure 6. Rich factor refers to the ratio of the numbers of differentially abundant metabolites annotated in this pathway to the numbers of all metabolites annotated to the same pathway. The greater the rich factor, the greater the pathway enrichment. The p value is another parameter for enrichment with a range from 0 to 1, the closer to 0, the more significance of the enrichment. As shown in Figure 6, biotin metabolism was significantly enriched, but the numbers of metabolites were much lower. Tryptophan metabolism was also significantly enriched with relatively higher metabolite numbers. Arginine and proline metabolism had a Rich factor of 0.33 with rather higher metabolite numbers. Lysine degradation had comparable Rich factors and also higher metabolite numbers. Additionally, beta-alanine metabolism was significantly relevant pathway of C. sinica larvae that were affected by ingestion of azadirachtin-treated P. xylostella larvae.  Table 1 shows some related metabolic pathways with representative differentially abundant metabolites and their upregulation and downregulation. Amino acids, carbohydrates, bile, lipids, membrane transports, cofactors and vitamins were primary metabolites that were affected in C. sinica larvae after ingestion of azadirachtin-treated P. xylostella larvae. The enrichment of pathways is presented in Figure 6. Rich factor refers to the ratio of the numbers of differentially abundant metabolites annotated in this pathway to the numbers of all metabolites annotated to the same pathway. The greater the rich factor, the greater the pathway enrichment. The p value is another parameter for enrichment with a range from 0 to 1, the closer to 0, the more significance of the enrichment. As shown in Figure 6, biotin metabolism was significantly enriched, but the numbers of metabolites were much lower. Tryptophan metabolism was also significantly enriched with relatively higher metabolite numbers. Arginine and proline metabolism had a Rich factor of 0.33 with rather higher metabolite numbers. Lysine degradation had comparable Rich factors and also higher metabolite numbers. Additionally, beta-alanine metabolism was significantly relevant pathway of C. sinica larvae that were affected by ingestion of azadirachtin-treated P. xylostella larvae. KEGG is the major public pathway-related database that includes not only genes but metabolites. † Pathway enrichment analysis identified significantly enriched metabolic pathways or signal transduction pathways in differential metabolites comparing with the whole background. The calculating formula is as follows: Here N is the number of all metabolites that with KEGG annotation, n is the number of differential metabolites in N, M is the number of all metabolites annotated to specific pathways, and m is number of differential metabolites in M. ‡ A variable importance in projection score of OPLS model was applied to rank the metabolites that best distinguished between two groups. § Means mass-to-charge ratio. Metabolites 2022, 12, x FOR PEER REVIEW 8 of 20 Figure 6. Metabolome map of significant metabolic pathways in C. sinica larvae affected by the ingestion of azadirachtin-treated P. xylostella larvae (pathway enrichment). Rich factor refers to the ratio of the number of annotated to this pathway in the differential metabolites to the number of annotated to this pathway in all metabolites. A larger rich factor indicates a higher degree of enrichment. p values range from 0 to 1, the closer to 0, the more significance of the enrichment. Figure 6. Metabolome map of significant metabolic pathways in C. sinica larvae affected by the ingestion of azadirachtin-treated P. xylostella larvae (pathway enrichment). Rich factor refers to the ratio of the number of annotated to this pathway in the differential metabolites to the number of annotated to this pathway in all metabolites. A larger rich factor indicates a higher degree of enrichment. p values range from 0 to 1, the closer to 0, the more significance of the enrichment.

Metabolite Changes Adversely Affected C. sinica Development
The present study showed that no mortality occurred in C. sinica larvae after consuming azadirachtin-treated P. xylostella larvae. However, their pupation and eclosion were significantly affected. Compared with the control treatment, 18% and 24% of the mature larvae were unable to perform pupation and eclosion, respectively (Figures 1 and 2). These results suggest that C. sinica larvae were able to obtain needed nutrients through digesting P. xylostella larvae to sustain their growth and even tolerate ingested azadirachtin, but the azadirachtin adversely affect C. sinica metabolism. As shown in Table 1 and Figure 6, over 10,000 metabolites across more than 20 pathways, including the metabolism of carbohydrates, lipid amino acids, vitamins and their cofactors, and amino acids were changed in C. sinica larvae.
Amino acids are fundamental for synthesizing proteins and phospholipids, energy production, and involved in morphogenetic processes. In this study, lysine degradation, tryptophan metabolism, phenylalanine metabolism, arginine and proline metabolism, valine, leucine, and isoleucine degradation were substantially down-regulated ( Table 1). The reduced metabolism of these amino acids could significantly impair C. sinica growth and development.
The carbohydrate metabolism is essential for cellular energy balance and for the biosynthesis of new cellular building blocks [43]. In this study, 10 carbohydrate pathways were down-regulated, and eight pathways were upregulated in both ESI+ and ESI− modes (Table 1). Among them, succinic acid was a differentially enriched metabolite in tricarboxylic acid (TCA) cycles. The TCA cycle, known as the citric acid cycle, has an important function that involves the intermediate compounds for the synthesis of amino acids and fatty acids and the formation of large quantities of adenosine triphosphate (ATP) that provides energy for various biological processes [24]. The downregulation of succinic acids of carbohydrate metabolites could cause a shortage of intermediate compounds and energy in azadirachtin-ingested C. sinica larvae, impairing their growth and development. Additionally, the amino sugar and nucleotide sugar metabolism pathway was enriched in azadirachtin-ingested C. sinica larvae. D-glycerate 3-phosphate was also among the differentially enriched metabolites in glycolysis/gluconeogenesis pathway. Glycolysis and gluconeogenesis are metabolic processes responsible for glucose degradation or glucose synthesis, respectively [44].
Ingestion of azadirachtin-treated P. xylostella larvae also affected the metabolism of vitamins and their cofactor in C. sinica larvae. Biotin is an essential substance for insects and affects the development of advanced larvae and pupae. The biotin metabolism pathway was significantly enriched in C. sinica larvae. The downregulation of this metabolite could adversely affect normal metabolism and C. sinica development.
Glycolysis is the main metabolic pathway of carbohydrates, such as galactose and fructose [44]. As an intermediate in both glycolysis and gluconeogenesis, the change in the relative content of D-glycerate 3-phosphate could also affect the generation of intermediate compounds and energy to maintain normal biological processes. Furthermore, the pathways of pentose and glucuronate interconversions, C-5 branched dibasic acid metabolism, ascorbate, and aldarate metabolism were enriched in azadirachtin-ingested C. sinica. Such a series of changes in carbohydrate metabolisms would affect the energy supply of C. sinica larvae, thus their development. Lipids are of vital importance to insects as energy sources and substrates for embryogenesis and development, pupation, metamorphosis, and other activities [45]. They are important components of insect cell membrane and also precursors of many insect pheromones [45]. Azadirachtin could influence the quantity and relative composition of fatty acids [30]. In the present study, 11 lipids or lipid-like metabolites were found to be differentially abundant, of which seven were upregulated, and four were down-regulated. The downregulated included palmitoleic acid, linolenic acid, 9-hydeoxyoctadecadienoic acid (9-OxoODE), and arachidic acid. Linolenic acid plays an important role in insect reproduction as it is a key constitute of oocyte dry mass and the major energy source for embryo development. The linolenic acid has been documented to be required for developing Heliothines subflexa [46]. The reduction in metabolism of linolenic acid in azadirachtin-ingested C. sinica larvae may potentially affect the pupation and eclosion, which required further confirmation.

Key Metabolic Pathways Affected by Feeding on Azadirachtin-Treated P. xylostella Larvae
The aforementioned analyses provide an overall spectrum of metabolite changes in C. sinica after ingestion of azadirachtin-treated P. xylostella larvae. The next question would be which metabolic pathways might be specifically implicated in the reduced percentages in pupation and eclosion C. sinica. Through the KEGG pathway analysis of the differentially abundant metabolites, the disturbed metabolic pathways caused by the consumption of azadirachtin-treated P. xylostella larvae were analyzed. A working model was constructed using the reference map deposited in the KEGG database (Figure 7). It was noticed that 20 differential metabolites were related to amino acid metabolic pathways. Among them, 15 metabolites were down-regulated. These results could imply that the ingestion of azadirachtin-treated P. xylostella larvae might result in the impairment in hydrolyzing proteins in C. sinica resulting in an insufficient supply of amino acids or directly affect amino acid metabolisms. Amino acids, particularly the essential ones are fundamental for insect growth and development [47]. A distinct biochemical characteristic of insects is their higher levels of free amino acids in the hemolymph [48], and the likely utilization of the free amino acids as silk protein synthesis to produce cocoon [49]. Furthermore, the eclosion is controlled by three peptide hormones: eclosion hormone, ecdysis-triggering hormone, and crustacean cardioactive peptide [50,51]. The reduction of free amino acids could hamper both pupation and eclosion of C. sinica.

Precautions When Azadirachtin and C. sinica Are Used in IPM
The changes in a wide range of metabolites in C. sinica larvae suggest that the action mode of azadirachtin is multifaceted with multiple biological targets. Thus, precautions should be taken when azadirachtin and C. sinica are to be used in IPM. Although this study was mainly focused on metabolites without further analysis of the biological influence of these molecular alterations on natural enemies, our data did document that C. sinica larvae after ingestion of larvae of P. xylostella treated with azadirachtin at 2.00 mg/L significantly reduced pupation and eclosion. Such adverse effects were related to the substantial changes in metabolomic profile of C. sinica. As mentioned previously, the premise for the use of pesticides, even botanical ones, in IPM is ascertaining their compatibility with beneficial predators. Our study showed azadirachtin and C. sinica are not compatible, which primarily agrees with the results of C. claveri in responses to azadirachtin [18][19][20][21]. Commercially, azadirachtin has been applied at much higher concentrations ranging from 5 to 100 mg/L [56]. Thus, detrimental effects to beneficial insects could be even severer. However, this does not exclude the use of azadirachtin and C. sinica at different time periods. Nevertheless, our study has raised the compatibility question between the two control tactics. Further studies are needed to identify specific mechanisms underlying the reduced pupation and eclosion of C. sinica larvae, effects on mating and oviposition of adults, and the relationship between azadirachtin concentrations and metabolite changes As shown in Figure 7, a majority of essential amino acid metabolisms were downregulated. A limited supply of tryptophan resulted in the decrease in the contents of indole, L-kynurenine, and 5-hydroxyindoleacetic acid. The downregulation of these compounds could limit the biosynthesis of acetyl-CoA, affecting the TCA cycle. It is worth mentioning that metabolites of the kynurenic pathway generally reach to peak concentrations in insects during pupation [52,53]. The reduced availability of kynurenine could impair pupation in C. sinica. Limited availability of lysine caused reduced biosynthesis of cadaverine and L-pipecolic acid. The downregulation of cadaverine also affected acetyl-CoA biosynthesis, subsequently affecting the TCA cycle. In the arginine metabolism pathway, low arginine in cells resulted in reduced synthesis of N-succinyl-L-glutamate, L-citrulline, and agmatine. Arginine can be converted to proline. Proline was a major substrate used in insect flight metabolism, which is known as the fuel of insect flight [54]. Choline was downregulated, which could indirectly affect the metabolism of serine and threonine. Pyruvate, an indirect derivative of choline, also from glycolysis is regarded as a key metabolite producing valine [55]. Valine also affects acetyl-CoA. L-valine was significantly low in C. sinica. Taken together, at the time of requiring higher levels of free amino acids in the hemolymph, the reduced amino acid metabolisms, particularly the essential ones significantly affected the pupation and eclosion of C. sinica.

Precautions When Azadirachtin and C. sinica Are Used in IPM
The changes in a wide range of metabolites in C. sinica larvae suggest that the action mode of azadirachtin is multifaceted with multiple biological targets. Thus, precautions should be taken when azadirachtin and C. sinica are to be used in IPM. Although this study was mainly focused on metabolites without further analysis of the biological influence of these molecular alterations on natural enemies, our data did document that C. sinica larvae after ingestion of larvae of P. xylostella treated with azadirachtin at 2.00 mg/L significantly reduced pupation and eclosion. Such adverse effects were related to the substantial changes in metabolomic profile of C. sinica. As mentioned previously, the premise for the use of pesticides, even botanical ones, in IPM is ascertaining their compatibility with beneficial predators. Our study showed azadirachtin and C. sinica are not compatible, which primarily agrees with the results of C. claveri in responses to azadirachtin [18][19][20][21]. Commercially, azadirachtin has been applied at much higher concentrations ranging from 5 to 100 mg/L [56]. Thus, detrimental effects to beneficial insects could be even severer. However, this does not exclude the use of azadirachtin and C. sinica at different time periods. Nevertheless, our study has raised the compatibility question between the two control tactics. Further studies are needed to identify specific mechanisms underlying the reduced pupation and eclosion of C. sinica larvae, effects on mating and oviposition of adults, and the relationship between azadirachtin concentrations and metabolite changes as well as threshold concentrations of azadirachtin to C. sinica for potentially better use of the two tactics for pest management.

Chrysopa sinica and Plutella xylostella
Chrysopa sinica and Plutella xylostella were raised in the insect rearing facility of the Key Laboratory of Natural Pesticide and Chemical Biology, Ministry of Education, South China Agricultural University, Guangzhou, China. C. sinica larvae were fed larvae of P. xylostella, and the adults were fed on a diet of P. xylostella larvae, along with 15% honey water, and yeast powder. The larvae of P. xylostella were fed leaves of the host cabbage (Brassica oleracea L.) plants, and the adults were fed 15% honey water. The temperature in the insect rearing facility was 25 ± 1 • C, relative humidity was 60 ± 5%, and a light-dark cycle of 16 h and 8 h.

Chemical Reagents and Instruments
Reagents including methanol, acetonitrile, ammonium acetate, ammonium hydroxide, and formic acid were purchased from CNW Technologies (ANPEL Laboratory Technologies, Inc. Shanghai, China). The internal standards of 2-chloro-L-phenylalanine was purchased from Shanghai Hengbai Biotech Co., Ltd. (Shanghai, China), and azadirachtin (>90%) were provided by Associate Professor Yongqing Tian at the South China Agricultural University.

Experimental Procedures and Samples Collection
Cabbage plants were singly grown in 15-cm pots filled with a peat-based potting substrate. When plants were at a stage of 10 leaves, 20 plants with a uniform growth size were randomly selected. The third instar larvae of P. xylostella were placed on cabbage leaves, 50 larvae per plant. Azadirachtin stock solution (10,000 µg/mL) was made by dissolving it in acetone. The stock solution was diluted with water resulting in a working solution of azadirachtin at 2.00 mg/L, which was sprayed on leaves of 10 plants (5 mL per plant) as azadirachtin treatment (T). The other 10 plants were sprayed with the same concentration of acetone (0.02%) in the same volume per plant as control treatment (CK). After the P. xylostella larvae had eaten the treated cabbage leaves for 12 h (Figure 1a), the third instar larvae of C. sinica were released and fed continuously with P. xylostella larvae from T and CK plants for 5 days, respectively (Figure 1b). The experiment was arranged as a complete randomized design with 10 replications. Five days later, more than fifty larvae of C. sinica were collected from each plant, quickly frozen in liquid nitrogen, and stored at −80 • C. The stored larvae samples were ground into a fine powder in liquid nitrogen and freeze-dried for 24 h until extraction.
To monitor the development of C. sinica, the pupation and eclosion of C. sinica larvae on cabbage plants of the two treatments were observed. After 10 days, when a complete white round cocoon was formed, the pupation was considered complete (Figure 1c). After 20 days, the adults broke out of the cocoon and spread their wings; they were deemed to have completed their emergence (Figure 1d). The experiment was arranged as a randomized complete block design with three replications (three blocks), each block had 50 larvae. The proportion of pupation and eclosion were calculated and analyzed using SPSS software platform (25.0) (IBM Corporation, Somers, NY, USA), and means were separated based on Tukey's HSD (honestly significant different) test at p < 0.01 and p < 0.001 levels.

Metabolites Extraction and Detection
The ground samples of control (CK) and azadirachtin treatment (T), 100 mg each, 10 replicates per treatment, were placed into Eppendorf tubes, respectively. To each tube was added 300 µL methanol and 20 µL 2-chloro-L-phenylalanine, the samples were vortexed for 30 s and sonicated for 5 min in the ice-water bath. The homogenate and sonicate circles were repeated for 3 times, followed by incubation at −20 • C for 1 h and centrifugation at 13,000 rpm and 4 • C for 15 min. The resulting supernatants were transferred to LC-MS vials and stored at −80 • C until analysis.
The quality control (QC) sample was prepared by mixing an equal aliquot of the supernatants from all of the samples to analyze the repeatability of samples under the same processing method. In the process of analysis, one quality control sample was inserted every 6-10 test analysis samples to monitor the repeatability of the analysis process.
The platform for LC-MS analysis consisted of an UHPLC system (1290, Agilent Technologies) with a UPLC HSS T3 column coupled to Q Exactive Orbitrap mass spectrometer (QEO MS) (Thermo Fisher Scientific). The supernatant (200 µL) was taken into the sample bottle (2 mL), respectively.
Mobile phase conditions were set as follows: the mobile phase A was 0.1% formic acid in water for positive, and 5 mmol/L ammonium acetate in water for negative (adjusted the PH value to 9.0 with ammonia), and the mobile phase B was acetonitrile. The elution gradient of the mobile phase was shown in supplementary file, Table S1.
Mass spectrometry conditions included the use of QEO MS to collect MS and MS/MS data with the electrospray (ESI) source conditions set as follows: spray voltage as 3800 V for positive or −3100 V for negative, capillary temperature 320 • C, sheath gas flow rate as 45 Arb, aux gas flow rate as 15Arb, full MS resolution as 70,000, MS/MS resolution as 17,500, strength of collision energy as 3, collision energy as 20/40/60 eV, scanning scope as 70-1000 m/z, scan rate as 7 Hz.

Data Preprocessing and Multivariate Statistical Analysis
The original LC-MS data files were converted to the mzML format by using ProteoWizard and processed by R package XCMS (version 3.2), including retention time alignment, peak detection, peak matching, and peak integration. The data were then filtered by the following criterion: sample numbers containing a metabolite were less than 50% of all sample numbers in a group. OSI-SMMS (version 1.0, Dalian Chem Data Solution Information Technology Co. Ltd, Dalian, China) was used for peak annotation after data processing with in-house MS/MS database.
To initially visualize the differences between different groups of samples, the principal component analysis (PCA) was applied. PCA analysis is an unsupervised multidimensional statistical analysis method that describes the characteristics of the original data set by compressing the original data into countless principal components, which can reflect the overall metabolic difference between each group of samples and the size of variation between the group samples. Partial least squares discriminant analysis (PLS-DA), as a supervised multivariate statistical analysis method, was used to distinguish the metabolomics profile of two groups by screening variables correlated to class memberships in which class memberships were coded in matrix form into Y [57]. Orthogonal projection to latent structures-discriminant analysis (OPLS-DA) is an extension of PLS-DA which incorporates an orthogonal signal correction (OSC) filter into a PLS model. The model quality was assessed based on cross-validation and permutation test [58]. The variable importance in projection (VIP) score of OPLS-DA model and t-test as a univariate analysis were applied to rank the metabolites that best distinguished the different groups in this study. Those with VIP ≥ 1 and a p-value of t-test < 0.05 were considered differential metabolites between two groups [59].

KEGG Pathway Analysis
After the metabolites were found, the metabolites were mapped to KEGG metabolic pathways for pathway analysis and enrichment analysis [60]. The main biochemical metabolic pathways and signal transduction pathways in differential metabolites were analyzed in this study. Significantly enriched metabolic pathways or signal transduction pathways in differential metabolites comparing with the whole background were identified through pathway enrichment analysis.

Conclusions
Azadirachtin as a botanical pesticide has been increasingly used for control of insect pests. This study investigated responses of predator C. sinica larvae after ingestion of P. xylostella larvae treated with azadirachtin, mainly at the metabolite levels. No mortality occurred in C. sinica larvae after consuming azadirachtin-treated P. xylostella larvae, but the percentages of pupation and eclosion of C. sinica were significantly reduced. Metabolomic analysis showed that azadirachtin has effects on the metabolism of amino acids, carbohydrates, lipid, cofactors and vitamins of C. sinica larvae. These effects may impair the growth and development of C. sinica, resulting in reduced pupation and eclosion percentages. Our studies for the first time documented substantial metabolite changes in C. sinica larvae after ingestion of azadirachtin-treated P. xylostella larvae and raise a question about the compatibility between azadirachtin and C. sinica in control of insect pests through IPM.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/metabo12020158/s1, Table S1: The elution gradient of the mobile phase in LC-MS analysis.