Glucosinolate induces transcriptomic and metabolic reprogramming in Helicoverpa armigera

Abstract

Glucosinolates protect plants from herbivory. Lepidopteran insects have developed resistance to glucosinolates which is well studied. However, the molecular effects of glucosinolate intake on insects are unexplored. To elucidate this, we performed transcriptomics and metabolomics of sinigrin-fed Helicoverpa armigera. Transcriptomics exhibits significant dysregulation of 2375 transcripts, of which 1575 are upregulated and 800 downregulated. Gene Ontology analysis of differentially expressed genes reveals that key hydrolases, oxidoreductases, and transferases are majorly affected. The negative impact of sinigrin is significant and localized in the endomembrane system and mitochondria. It also disturbs various biological processes such as regulation of protein metabolism and cytoskeletal organization. Furthermore, H. armigera putative myrosinase-like enzymes may catalyze the breakdown of sinigrin to allyl isothiocyanate (AITC). AITC targets the electron transport chain causing oxidative stress. KEGG pathway enrichment shows significant upregulation of oxidative phosphorylation, glutathione metabolism and amino acid metabolism. Activation of these pathways induces glutathione synthesis for sinigrin detoxification. Differential gene expression indicates upregulation of glutathione S-transferase and succinate dehydrogenase suggesting mitochondrial impact. Transcriptomics data correlated with metabolomics show changes in serine, methionine, ornithine, and other metabolite levels. It corroborates well with the transcript alterations supporting the increased glutathione production. Thus, our data suggest that sinigrin generates oxidative stress in H. armigera and insects alter their metabolic wiring to overcome sinigrin-mediated deleterious effects.

Supplementary Information

The online version contains supplementary material available at 10.1007/s13205-020-02596-5.

Introduction

Plants produce an array of secondary metabolites to deter herbivorous insects (Zunjarrao et al. 2019). Glucosinolates (GLS) are one of the primary defence secondary metabolites. They are sulphur-rich, anionic glucosides derived from amino acids, and stored in plant S-cells (Halkier and Gershenzon 2006). Upon damage caused by insect feeding, myrosinases hydrolyze them into unstable aglucone moiety. Aglucone moiety rearranges into toxic compounds such as isothiocyanates (ITCs), oxazolidine-2-thione, nitriles, epithonitriles, and thiocyanates (Wittstock et al. 2003; Jeschke et al. 2015). Various effects of GLS are well documented in generalist and specialist herbivores (Wittstock et al. 2003; Arany et al. 2008; Müller et al. 2010; Ali and Agrawal 2012; Kos et al. 2012; Jeschke et al. 2015, 2016, 2017). Various specialist herbivores have developed novel strategies to counteract the plant GLS (Ahn et al. 2019; Chen et al. 2020; Friedrichs et al. 2020; Liu et al. 2020; Malka et al. 2020; Shukla and Beran 2020; Sun et al. 2020; Yang et al. 2020).

GLS-derived ITCs caused developmental retardation, deformities, and death in cotton bollworm (Helicoverpa armigera: Noctuidae) and other lepidopteran insects (Bhushan et al. 2016; Jeschke et al. 2017; Agnihotri et al. 2018). However, mechanisms involved in this multifactorial negative effect on insect physiology are still enigmatic. In African cotton leafworm (Spodoptera littoralis: Noctuidae), ITC feeding elevated protein catabolism (Jeschke et al. 2016). In Maize weevil (Sithophilus zeamais: Curculionidae) and Indian meal moth (Plodia interpunctella: Pyralidae), ITC treatment inhibited mitochondrial respiratory complexes affecting cellular respiration (Mansour et al. 2012; Zhang et al. 2016; Hou et al. 2016). All these reports highlighted ITC-mediated impairment in the mitochondrial functions. Therefore, mitochondria can be the primary target of GLS-derived ITCs, and that affects the central metabolism of insects.

Previously, H. armigera larvae showed developmental delay upon sinigrin feeding (Agnihotri et al. 2018). To further dissect its adverse effects on insect physiology, we fed H. armigera larvae with an artificial diet containing sinigrin. We used whole insect tissue of control and sinigrin-fed 5th instar larvae for the de novo transcriptomics. Differential expression of significantly dysregulated genes is validated using quantitative real-time PCR. Targeted metabolomics of treated insect was performed to enrich the transcriptomics data. We anticipate that this study will shed light on the mode of action of sinigrin on generalist herbivores.

Materials and methods

Insect culture and bio-assay

Insects were procured from the National Bureau of Agricultural Insect Resources, Bengaluru, India. They were grown and maintained in the lab under controlled conditions. Second generation insects were used for the feeding and -omics experiments. The larvae were maintained on the chickpea-based artificial diet (Supplementary Data 1). Plastic containers with screw caps were used to grow insects. A total of 30 control and 30 sinigrin-treated insects were maintained for 10 days. Humidity (80% relative humidity), photoperiod (16 h light, 8 h dark), and temperature (28 °C) were maintained as described earlier (Joshi et al. 2014). Bio-assay was conducted by feeding 2nd instar H. armigera larvae on an artificial diet containing 250 ppm of sinigrin hydrate (Sigma) and control feed without sinigrin hydrate. The diet was prepared as described previously (Nagarkatti and Prakash 1974) (Supplementary Data 1). The concentration of sinigrin was determined considering the survivability and developmental deformities observed in H. armigera in the previous study (Agnihotri et al. 2018). Bio-assay was performed till insects reached 5th instar, after which insects were flash-frozen using liquid nitrogen and stored at − 80 °C until further analysis.

RNA sequencing, de novo assembly, and differential expression analysis

RNA was extracted from three 5th instar control and three 5th instar sinigrin-treated insects using TriReagent (Invitrogen). Insects were ground to a fine powder in a pre-cooled mortar pestle using liquid nitrogen. 100 mg powder was weighed to which 1 ml of TriReagent was added, and samples were allowed to stand at room temperature (RT) for 5 min. To this 0.2 ml of chloroform was added for phase separation and samples were vortexed for 15 s. Samples were then centrifuged at 12,000 × g for 15 min at 4 °C. The colourless aqueous phase was collected in a separate tube. To this 0.5 ml ice-cold isopropanol was added and incubated at RT for 10 min. Samples were then centrifuged at 12,000 × g for 10 min at 4 °C. The supernatant was discarded and to the RNA pellet 1 ml of 75% ice-cold ethanol was added. Samples were vortexed, followed by centrifugation at 7500 × g for 5 min at 4 °C. Ethanol was removed and the RNA pellet was air-dried for 10 min. The RNA pellet dissolved in 30 μl of nuclease-free water. RNA-Seq library was prepared as per Illumina-compatible NEB NextUltra^™ Directional RNA Library Prep protocol. Illumina HiSeq sequencer was used to obtain 150 bp paired-end (PE) reads. Raw reads were processed for the quality assessment using FastQC. Adaptor removal and filtering of low-quality reads were achieved using Cutadapt (Martin 2011). De novo assembly of processed reads was done by Trinity (default k-mer) (Grabherr et al. 2011). Assembled transcripts were clustered using CD-HIT-EST with 95% similarity between the sequences to eliminate redundant sequences. DESeq was used to identify differentially expressed genes (Anders and Huber 2010). Sequencing (Uneven library size/depth) bias among the samples was removed by library normalization using size factor calculation in DESeq. DESeq normalized expression values (RPKM) were used to calculate fold change. Based on these library normalized expression values, transcripts with log2fold change greater than 1 were (DEG) considered as upregulated and lesser than − 1 were deemed to be downregulated (p value < 0.01).

DEG functional classification, GO enrichment and pathway analysis

Clustered transcripts were annotated using BLAST against lepidopteran proteins from Uniprot (419,418) (Altschul et al. 1990). DEGs (UniProt IDs) were sent to KOBAS for gene ontology (GO) and KEGG pathway analysis (Figs. 1, 2) (Wu et al. 2006).

Fig. 1.

The figure displays a volcano plot of 53,773 transcripts that were detected in the control and sinigrin-treated insects. Red colour indicates downregulated (log2foldChange <  − 1) transcripts, green colour for upregulated (log2foldChange > 1) and blue colour for neutrally regulated ones.The total number of downregulated transcripts is 13567, that of upregulated is 12606, and that of neutrally regulated is 27600

Fig. 2.

a Figure depicts GO enrichment analysis for DEGs. Green coloured are upregulated and red coloured are downregulated genes in the given tornado plots. In the case of molecular function, various enzyme activities along with protein and ion binding are impacted. The cellular components affected are plasma membrane, nucleus, and organelle membranes. Whereas, the biological processes affected are nucleoside metabolism, lipid metabolism, and carbohydrate metabolism. b Figure indicates KEGG Pathway Analysis. Tornado plot shows the pathway category on the right and sub-category on the left-hand side of the plot. The most impacted pathway is glutathione metabolism, followed by oxidative phosphorylation, and protein processing in the endoplasmic reticulum

Homology modelling and docking

Seven unique putative myrosinases were modelled using SWISS-MODEL server (Waterhouse et al. 2018). The details of the template and model parameters are given in Table 1. The models were further energy minimized in Swiss PDB Viewer 4.10 (http://www.expasy.org/spdbv/) and subjected to docking in Autodockvina (Trott and Olson, Trott and Olson 2010). A library of 35 glucosinolates (reported in KEGG) and four control ligands with known binding to glucosidases (acarbose, raffinose, amygdalin, oseltamivir) were considered for docking. Furthermore, for comparison PDB structure of aphid myrosinase (PDB ID: 1WCG) was subjected to the same docking procedures (Husebye et al. 2005). Active site-specific docking was carried out for binding energy calculations. Best-docked complexes (based on affinity values) were visualized in the Biovia Discovery Studio 4.5 (DassaultSystèmes BIOVIA, Discovery Studio Modeling Environment, Release 2017, San Diego: DassaultSystèmes, 2016) (Fig. 3).

Table 1.

Details of the template and model parameters

Protein	Template	% Coverage	% identity	RMSD
XP_021196646.1 (myrosinase_1)	5CG0.1.A	96	46.41%	0.157
XP_021191418.1 (myrosinase_2)	5CG0.1.A	99	71.87%	0.118
XP_021191417.1 (myrosinase_3)	5CG0.1.A	99	71.66%	0.115
XP_021191416.1 (myrosinase_4)	5CG0.1.A	99	78.60%	0.101
XP_021197062.1 (myrosinase_5)	5CG0.1.A	96	45.78%	0.204
XP_021196645.1 (myrosinase_6)	5CG0.1.A	96	46.81%	0.183
XP_021194605.1 (myrosinase_7)	3VIK.1.A	97	51.80%	0.148

Fig. 3.

a Log2RPKM of the seven myrosinase-like enzymes that show high expression in sinigrin-treated insects b Clustered heatmap of z-scores of binding affinities obtained in docking. Myrosinase_6 and aphid myrosinase clustered together, offering a similar binding pattern. Myrosinase_7 is closely connected to myrosinase_6. Myrosinase_2, 3 and 4 form a single cluster, while myrosinase_1 and 5 clade out separately. This pattern depicts the diversity in myrosinase-like enzymes of H. armigera c Interaction maps of 3 best docked H. armigera myrosinases with sinigrin. Green bonds conventional H-bonds, orange indicate Pi-anion interactions, pink symbolize Pi-alkyl interactions. Red indicates unfavourable interactions. Dark pink indicate Pi-Pi T shaped and purple indicate Pi-sigma interactions

Validation of RNA-Seq by qRT-PCR

RNA was extracted from three random control and sinigrin-treated insects using TRIReagent as described in an earlier section. The genomic DNA was removed using RQ1 RNase-free DNase (Promega). 2 μg RNA was reverse-transcribed into cDNA with high-capacity cDNA reverse transcription kit (Applied Biosystems). Quantitative real-time PCR (qRT-PCR) was performed using TB Green Premix Ex Taq II (Tli RNase H Plus) (TaKaRa) with 7900HT Fast Real-Time PCR System with 384-Well Block Module (Applied Biosystems^™). The reaction condition was set as follows: 95 °C for 30 s, followed by 40 cycles of 95 °C for 5 s, 55 °C for the 30 s followed by a dissociation curve. The fluorescence signal was monitored automatically in each cycle. Relative expression levels of specific mRNAs were measured with the 2^(−ΔΔCt) analysis method. The expression values were normalized using the elongation factor (EF) gene, that showed identical expression in control and sinigrin-treated samples (Zhang et al. 2015). For each biological replicate (each insect), three technical replicates were analysed. p value was calculated using student’s T test in MS Excel. Primers used in this validation are listed in Supplementary Data 2.

Metabolite extraction and analysis using LC-QTOF-MS

Metabolite extraction was done using 100 mg of crushed tissue of four 5th instar insect larvae. To the 100 mg powder, 100% MS grade methanol was added for metabolite extraction. This suspension was first vortexed for 10 min at room temperature, then sonicated for 20 min and finally centrifuged at 18000 × g for 10 min. The supernatant was filtered through a 0.2 µm syringe filter and the filtrate was transferred to sample vials. The LC-QTOF-MS analysis was carried out on Agilent 6530 Q-TOF (Agilent, USA) mass spectrometer connected to HPLC Prime Infinity II 1260 system (800 bar) and dual electrospray ionization (ESI) source was used for ionization. For LC-based metabolite separation, the Infinity Lab Poroshell 120 EC-C18 (2.1 × 150 mm, 1.9 μm particle size, Agilent, USA) column was used at 40 °C with a flow rate of 0.3 ml/min. The metabolites were separated using a 20 min gradient with 100% MS grade water (Solvent A) and 100% MS grade acetonitrile (Solvent B) both containing 0.1% formic acid. The LC method started with 2% B for the first 0.3 min and increased to 30% in the next 2 min. The B% was increased from 30 to 45% till 7 min and further increased to 98% to 12 min at which it was held for the next 3 min. The column was equilibrated to the initial ratio of solvents (98% A: 2% B) in the last 5 min. The real-time mass correction was done with two masses (121.0508 and 1221.9906) using dedicated input needle. The MS data from five independent biological replicates were acquired in 2 GHz extended dynamic range. The MS parameters tuned as; gas temperature 325 °C, drying gas 10 L/min, and nebulizer at 35 psi and fragmentor 120 V. Non-targeted metabolite data analysis, including deisotoping/deconvolution, peak picking, and retention time correction, were performed using XCMS online web server (https://xcmsonline.scripps.edu) (Tautenhahn et al. 2012). The peak lists obtained with RT, m/z and respective peak area intensities were grouped by principal component analysis (PCA) in unit-variance scale. In the case of targeted metabolite analysis, the local database of compounds was created using Agilent’s PCDL B.08.00 tool. The identification of molecules was based on accurate mass, isotopic fidelity and isotope distribution. The peak area of metabolites was calculated using Agilent MassHunter Qualitative Navigator B.08.00 and Qualitative Workflow B.08.00 tools to calculated relative abundance and fold change.

Results and discussion

Glucosinolate ingestion leads to transcriptional alteration as a molecular response

Transcriptome sequencing yielded 44.85 and 41.21 million raw reads (150 × 2) from control and sinigrin-treated insects, respectively (Bioproject No.: PRJNA673794). We processed and assembled around 43 and 39 million high-quality reads sample wise. The clustering of assembled sequences produced 39,526 control and 33,282 sinigrin-treated transcripts. We detected a total of 53,773 transcripts (Fig. 1). DEG analysis provided 12,606 upregulated and 13,567 downregulated transcripts in sinigrin-treated insects (Supplementary Data 3). The top upregulated (11–12 log2 folds) genes were Glucose dehydrogenase [FAD, quinone]-like, glutathione S-transferase-like gene, juvenile hormone-suppressible protein, lipase, and serine protease inhibitor. A disintegrin and metalloproteinase with thrombospondin motif (ADAMT), luciferin 4-monooxygenase-like, and lysophospholipid acyltransferase 7-like gene were the top downregulated genes (7–8 log2 folds) (Fig. 4a) (Supplementary Data 4). The 2375 transcripts (1575 upregulated, 800 downregulated) were submitted for GO and KEGG pathway analysis. GO enrichment for molecular function showed that enzyme expression (hydrolase, oxidoreductase, transferase, and ligase) was deeply affected (Fig. 2a). Majorly involved biological processes were nucleotide phosphate, lipid and carbohydrate derivative processes, regulation of protein metabolism, cytoskeletal organization, hormone and sterol metabolism, and endocytosis. The cellular component analysis showed significant changes in the endomembrane system and mitochondria. KEGG pathway analysis gave major hints on the action of sinigrin on oxidative phosphorylation and glutathione (GSH) metabolism, followed by amino acids metabolism (Fig. 2b).

Fig. 4.

a Heat map depicting highly dysregulated transcripts selected from transcriptome based on logRPKM values b Expression (negative deltaCt) values of selected genes c Fold change (2^-ddCt) of selected genes at 0.05LOS. Succinate dehydrogenase, a key enzyme of oxidative phosphorylation is observed to be almost 10 folds upregulated. Along, with that GST is also upregulated

Mitochondria as a site of glucosinolate action

AITC, the breakdown products of sinigrin act as uncouplers of oxidative phosphorylation. This results in electron transport without the formation of ATPs, increased rate of respiration, and CO₂ emission. This forces insects to utilize the stored energy reserves, finally leading to starvation and death (Tsao et al. 2002). Upon AITC fumigation, insects exhibited increased CO₂ emission along with vacuolization of the mitochondrial matrix and reduced the number of cristae (Mansour et al. 2012). Transcriptomics of AITC fumigated S. zeamis revealed cytoskeletal collapse and mitochondrial dysfunction (Zhang et al. 2017). Also, in vitro and in vivo studies showed that AITC inhibited NADH: ubiquinone oxidoreductase and cytochrome c oxidase (Zhang et al. 2016). With the help of biochemical and in silico studies, it was concluded that COX-II, the subunit of cytochrome c oxidase is the site of action of AITC (Hou et al. 2016). ITCs also inhibited several dehydrogenases, including succinate and glucose dehydrogenase (Miko and Chance 1975). In congruence to the prior literature, our transcriptomics of sinigrin-treated insects showed upregulation of the oxidative phosphorylation genes (NADH dehydrogenase, F-type ATPase, cytochrome bc1 complex, cytochrome c oxidase, and succinate dehydrogenase) (Table 2). The real-time analysis of succinate dehydrogenase and glucose dehydrogenase showed 3.2 and 1 log2 fold upregulation, to compensate for sinigrin-mediated inhibition (Fig. 4b). Therefore, we speculate the upregulation of oxidative phosphorylation as a trigger of sinigrin-mediated toxicity.

Table 2.

Upregulated genes of ETC

Sr. No	Transcript ID	Control	Singrin	Log2 FC	Annotation
1	Master_Transcript_52696	1.7323	83.1227	5.5844	NADH dehydrogenase
2	Master_Transcript_48136	8.6618	247.0591	4.8340
3	Master_Transcript_43086	6.9295	108.5213	3.9690
4	Master_Transcript_52753	2.5985	96.9765	5.2218
5	Master_Transcript_43355	3.4647	200.8798	5.8574
6	Master_Transcript_41288	1.7323	131.6109	6.2473
7	Master_Transcript_51878	31.1828	1183.3442	5.2459	Succinate dehydrogenase
8	Master_Transcript_52525	3.4647	130.4564	5.2346	Cytochrome bc1
9	Master_Transcript_50044	1.7323	46.1792	4.7364
10	Master_Transcript_45887	8.6618	375.2067	5.4368	Cytochrome c oxidase
11	Master_Transcript_43401	6.9295	252.8315	5.1892	ATP synthase
12	Master_Transcript_51832	5.1971	86.5861	4.0583
13	Master_Transcript_43338	3.4647	136.2288	5.2971

Sinigrin-mediated dysregulation of mitochondria is also related to juvenile hormone (JH). JH suppressible protein is upregulated in transcriptomics and validated in real-time PCR (Fig. 4, Supplementary Data 3). JH is an acyclic sesquiterpenoid which regulates insect development, reproduction, and diapause (Jindra et al. 2013). The primary function of JH is to prevent metamorphosis allowing larval growth. However, similar to AITC, JH is observed to act as an uncoupler of oxidative phosphorylation (Chefurka 1978). In fruit flies (Drosophila sp.: Drosophilidae) application of JH and its analogues resulted in mitochondrial swelling (Farka and Ut’áková G 2001). Therefore, even though JH is indispensable for insect development, it may adversely affect the mitochondria. Hence, to prevent these effects in an already stressed cellular environment, JH suppressible protein must be upregulated.

Sinigrin causes induction of detoxification mechanisms

Insects employ maximum resources to curb the sinigrin-mediated oxidative stress. Detoxification of plant secondary metabolites occurs through enzymes such as CYP450s, CCEs, GSTs and UGTs (Schweizer et al. 2017). Most of these genes upregulated in H. armigera post-sinigrin-feeding. The critical process of sinigrin detoxification is via glutathione (GSH) metabolism (Schramm et al. 2012). KEGG pathway analysis showed upregulation of GSH metabolism (Fig. 2b). Glutathione-s-transferase (GST) conjugates ITCs to GSH in the insect midgut, which subsequently excreted via faeces (Jeschke et al. 2016). GST upregulated in the transcriptome (3–4 log2 folds) and real-time PCR (2.18 log2 folds) upon sinigrin treatment (Fig. 4, Supplementary Data 4).

Sinigrin showed strong binding and multiple interactions with H. armigera myrosinase-like enzymes

In our previous study, we observed that sinigrin was converted to AITC when incubated with H. armigera protein extract. Therefore, we speculated the presence of myrosinase-like enzymes in H. armigera (Agnihotri et al. 2018). Interestingly, we observed seven unique variants of myrosinase-like enzymes upregulated in the current transcriptome (Fig. 3a, Supplementary Data 5). Modelling and molecular docking gave critical insights on myrosinase binding to glucosinolates (Supplementary Data 4). Glucosinolates derived from methionine are more diverse than those derived from other aliphatic or aromatic amino acids (Supplementary Data 6). Amygdalin, raffinose are known substrates and acarbose, oseltamivir are inhibitors of beta-glucosidase. Amygdalin and acarbose showed better binding than the average affinity of GLS to myrosinases, except for myrosinase_1. Raffinose showed comparable binding, similar to other GLS against all myrosinases. Oseltamivir showed efficient but lower than average GLS binding for any myrosinase. Myrosinase_6 showed the strongest binding with N-Sulfoindol-3-yl-methylglucosinolate (− 9.9 kcal/mol), and weakest was that of myrosinase_5 with Glucoputranjivin (− 5.2 kcal/mol) (Supplementary Data 5). H. armigera and aphid myrosinases bind more strongly to aromatic GLS particularly indolic. Myrosinase_5 showed comparatively better binding to GLS from methionine and leucine. Myrosinase_3, 4 and 6 showed strong binding to sinigrin (Fig. 3c). In clustering analysis of standardized docking scores (z-scores), myrosinase_6 and aphid myrosinase clustered together showing a similar binding pattern. Myrosinase_2, 3 and 4 make one cluster, while myrosinase_1 and 5 clade out separately (Fig. 3b). This clustering based on docking scores depicts that H. armigera harbours diversity in myrosinase-like enzymes, in terms of sequence and activity (Supplement 4, 5).

In nature, when insects feed on plants, plant myrosinase breakdown GLS to ITCs. Though myrosinase is a plant enzyme, few insects have evolved their myrosinases (Pontoppidan et al. 2001; Jones et al. 2002; Beran et al. 2014). Myrosinase from cabbage aphid (Brevicoryne brassicae: Aphididae) is well characterized (Francis et al. 2002). They use their myrosinase for defence against their predators. This myrosinase degraded gluconapin and sinigrin (methionine-derived GLS). However, enzyme kinetics showed that aromatic GLS have higher affinity than aliphatic (Francis et al. 2002). Striped flea beetle (Phyllotreta striolata: Chrysomelidae) feeding on Brassica plants showed sequestration of GLS, followed by their hydrolysis and emission of ITCs (Beran et al. 2014). However, myrosinase is not yet reported in lepidopteran insects. Thus, we suspect that these putative myrosinases facilitate H. armigera to survive on a wide range of hosts, including brassicaceae family. H. armigera is affected by the presence of aliphatic GLS and not so much by indolic (Zalucki et al. 2017). Thus, our docking study and literature suggests that H. armigera is more resistant to indolic GLS than to aliphatic. The fate of the degradation of indolic GLS remains to be studied.

Global metabolomic alterations occur in sinigrin-fed insects to assist detoxification machinery

Untargeted metabolomics detected 4500 metabolites depicting insect’s metabolic changes in response to sinigrin (Fig. 5a). Of these, about 1200 metabolites showed significant dysregulation with p value < 0.01 and fold change > 1.5. Most of them belonged to glutathione-mediated detoxification, TCA cycle, amino acid metabolism, leukotriene biosynthesis, and the urea cycle. Targeted metabolomics was carried out for selected metabolites of these pathways to get insights into the precise metabolic changes (Table 3). The top upregulated metabolites were serine, glutathione disulfide (GSSG) and asparagine. Whereas phenylalanine, ornithine and isoleucine show a substantial decrease in their levels (Fig. 5b).

Fig. 5.

a PCA analysis of untargeted metabolomics. The PCA is calculated using the feature intensities from all the samples. Red indicates control insects whereas blue indicates sinigrin-treated insects. b Box-whisker plot of top-up- and downregulated metabolites. Serine is the most upregulated (2.48 folds) metabolite. Ornithine is the most downregulated metabolite (2.5 folds)

Table 3.

Targeted metabolites analysis with the details of mass spectrometry data along with fold change (CONTROL/SINIGRIN)

Sr. No	Mean RT (min)	Molecular formula	Ion annotation	Neutral mass (Da)	Observed mass (Da)	Mass error in ppm	Description	Fold change
1	5.33	C8H8O3	[M + H]+	152.0474	153.0547	0.232	3,4-dihydroxyphenylacetaldehyde	0.78
2	1.49	C5H7NO3	[M + H]+	129.0428	130.0499	0.744	5-oxoproline	1.54
3	1.01	C6H14N4O2	[M + H]+	174.1119	175.1191	0.933	Arginine	0.89
4	1.175	C4H8N2O3	[M + H]+	132.0538	133.0608	− 2.77	Asparagine	1.94
5	1.50	C4H5N3O	[M + H]+	111.0432	112.0505	− 0.80	Cytosine	0.93
6	1.071	C5H10N2O3	[M + K]+	146.0694	185.0333	− 3.63	Glutamine	0.59
7	1.50	C10H17N3O6S	[M + H]+	307.0837	308.0907	0.237	Glutathione	1.15
8	1.60	C20H32N6O12S2	[M + H]+	612.1526	613.1598	0.718	Glutathione disulfide	1.81
9	11.90	C16H32O2	[M + NH4]+	256.2407	274.2746	0.629	Hexadecanoic acid	0.67
10	1.04	C6H9N3O2	[M + H]+	155.069	156.0761	− 1.797	Histidine	1.42
11	1.93	C6H13NO2	[M + H]+	131.0942	132.1015	− 1.196	Isoleucine	0.31
12	2.10	C6H13NO2	[M + H]+	131.0942	132.1017	− 0.675	Leucine	1.23
13	1.25	C6H14N2O2	[M + K]+	146.1051	185.0685	− 1.88	Lysine	0.80
14	1.51	C5H11NO2S	[M + H]+	149.0507	150.0579	− 1.69	Methionine	0.90
15	1.0007	C6H11NO2	[M + H]+	129.0784	130.0861	− 1.961	N4-acetylaminobutanal	0.80
16	14.41	C18H34O2	[M + NH4]+	282.2564	300.2897	1.682	Octadecenoic acid	0.89
17	1.0028	C5H12N2O2	[M + H]+	132.0897	133.0969	− 0.818	Ornithine	0.39
18	3.5004	C9H11NO2	[M + H]+	165.0783	166.0855	− 1.764	Phenylalanine	0.21
19	1.1119	C5H9NO2	[M + NH4]+	115.0631	133.0974	− 0.305	Proline	0.79
20	1.0739	C3H7NO3	[M + H]+	105.0425	106.0498	0.902	Serine	2.48
21	0.9585	C7H19N3	[M + H]+	145.158	146.1653	0.403	Spermidine	1.07
22	9.474	C14H28O2	[M + NH4]+	228.2089	246.2428	0.522	Tetradecanoic acid	0.98
23	1.0886	C4H9NO3	[M + H]+	119.0581	120.0653	− 2.324	Threonine	1.11
24	4.0403	C11H12N2O2	[M + H]+	204.0899	205.0972	0.251	Tryptophan	0.42
25	1.8062	C9H11NO3	[M + H]+	181.0733	182.0807	− 1.935	Tyrosine	0.61
26	1.5072	C5H11NO2	[M + H]+	117.0783	118.0856	− 2.85	Valine	0.52

Induction of oxidative stress and metabolic flux towards GSH synthesis

GSH plays a pivotal role in the detoxification of deleterious substances that disturb the redox balance of the cells (Ahlawat et al. 2020). GSH metabolism is the primary pathway of sinigrin detoxification. However, this detoxification capacity depends on GSH availability. The depletion of GSH has detrimental effects on intracellular homeostasis (Ahmad 1992; Felton and Summers 1995). In our study, the prolonged exposure of insects to the sinigrin diet must have elevated the GSH demand. Moreover, we observed a change in glutathione redox ratio (GSH: GSSG) and increased GSSG levels, an indicator of oxidative stress (Fig. 5b). Oxidative stress reduces the glutathione redox ratio (Jones 2002). Metabolite quantification indicated a reduction in glutathione redox ratio from 39.88 to 25.39 in sinigrin-fed insects. Also, treated larvae showed overexpression of glutathione peroxidase transcripts, needed conversion of GSH to GSSG (Kiriyama et al. 2013). This decrease in the glutathione redox ratio may marked for the oxidative stress.

GSH is synthesized from three different amino acids: glutamate, cysteine, and glycine. Of these, cysteine is the rate-limiting substrate (Aoyama et al. 2008). ITC-treated S. littoralis shows the reduction in cysteine and methionine levels due to their rapid utilization in GSH metabolism (Métayer et al. 2008; Jeschke et al. 2015). Thus, we attribute the absence of cysteine and reduced methionine levels (0.9 fold decrease) in metabolomics to the elevated need for GSH (Supplementary Data 6). Moreover, as a result of the oxidative stress, the one-carbon cycle shifts towards transsulfuration from methylation, to produce cysteine for GSH synthesis (Deplancke and Gaskins 2002). Mainly, serine along with methionine makes cystathionine, via the transsulfuration pathway. Cystathionine broke down to free cysteine via upregulated cystathionine gamma-lyase (Master_Transcript_44501). Hence, post-sinigrin-treatment serine can maintain GSH levels. Therefore, to amend these changes and to maintain methylation potential of an organism a high concentration of serine (2.48 fold upregulation) is utmost necessary (Fig. 5b) (Davis et al. 2004; Maddocks et al. 2016).

In contrast to serine, ornithine is reduced by 2.5 folds, post-sinigrin treatment, suggesting its increased utilization (Fig. 5b). Ornithine is involved in the urea cycle, arginine, proline and GSH metabolism. Ornithine decarboxylase (Master_Transcript_48188) that converts ornithine to putrescine is upregulated (Morris and Pardee 1966). The downstream products of putrescine are polyamines (spermidine and spermine). Both spermidine and spermine are involved in the protection of mitochondria against oxidative stress and are a part of GSH metabolism (Rigobello et al. 1993). The protection given by these polyamines is independent of GSH-mediated protection (Rider et al. 2007). Putrescine also plays a role in diapause induction and development in H. armigera and other insects. Sinigrin-induced mitochondrial dysregulation may trigger the conversion of ornithine to putrescine (Cayre et al. 1997; Mitsuhashi 1998; Wu et al. 2010).

We also observed 1.5 times elevation in 5-oxoproline levels in sinigrin-fed insects (Supplementary Data 6). 5-oxoproline is a natural precursor of glutamate and a part of GSH metabolism (Kumar and Bachhawat 2012). 5-oxoproline is produced from GSH; however, it is less likely here as GSH might be utilized for sinigrin detoxification (Van der Werf et al. 1971). Another pathway of 5-oxoproline production is from glutamate (Krishnaswamy et al. 1960; Orlowski et al. 1969; Seddon et al. 1989). During the first step of GSH biosynthesis, glutamate activated by γ-Glutamylcysteine synthase (γ-GCS). Activated glutamate binds to cysteine to form γ-glutamylcysteine (γ-GC) (Franklin et al. 2009). However, during GSH deficiency, γ-GC is converted to 5-oxoproline by γ-Glutamylcyclotransferase. Thus, we observe accumulation of 5-oxoproline.

Elevated protein catabolism and lipid synthesis to alleviate sinigrin toxicity

Excess of ITCs and its detoxification limited by cysteine levels impose physiological stress on the insect (Jeschke et al. 2016). This discrepancy is compensated by raised levels of protein catabolism to supply amino acids (Etebari et al. 2007; Nath et al. 1997; Jeschke et al. 2016). Our transcriptomics highlighted the biological processes such as digestion and positive regulation of proteolysis. We also observed upregulation of Cathepsin B (transcriptomics and real-time PCR) (Fig. 4). Oxidative stress leads to the overexpression of Cathepsin B and Cathepsin D in Bombyx mori (Saikhedkar et al. 2015). Cathepsin B is a lysosomal cysteine protease and ITCs inhibited cysteine proteases (Tang and Tang 1976). Sinigrin inhibited cathepsin activity in H. armigera (Agnihotri et al. 2018). Cathepsin B functions in intracellular protein catabolism (Mort and Buttle 1997). Therefore, to compensate for the ITC inhibition, cathepsin B is upregulated and in turn, aids in protein catabolism for increased GSH synthesis.

In the case of oxidative stress, lipids undergo peroxidation (Mansour and Mossa 2009). The free radicals capture electrons from cellular membrane lipids, which deplete the lipid content. In our transcriptome, we see upregulated glycerolipid metabolism (Fig. 2b). It results in complete or partial lipid hydrolysis to yield monoacylglycerol, glycerol and fatty acids, which are utilized in energy metabolism. Products of the glycerolipid pathway enter the glycerophospholipid metabolism, which is also upregulated. The downstream products of this pathway enter various metabolic pathways like glycine, serine and threonine, which are essential for GSH synthesis. We observed downregulation of lysophospholipid acyltransferase 7-like (LPLATs) in the transcriptome (7 log2 folds) and real-time PCR as well (Fig. 4, Supplementary Data 3). LPLATs are involved in both glycerolipid and glycerophospholipid metabolism. LPLATs catalyze the re-acylation step of the phospholipid remodelling pathway (Yamashita et al. 1997).

Further, we observed triacylglycerol lipase to be upregulated (11 log2 folds) in transcriptomics (Supplementary Data 3). Lipases hydrolyze lipids and are a part of glycerolipid metabolism. In tobacco budworm (Heliothis virescens: Noctuidae) lipase was upregulated upon GLS feeding (Schweizer et al. 2017). In South American palm weevil (Rhynchophorus palmarum: Curculionidae) and tobacco hornworm (Manduca sexta: Sphingidae), lipase activity increased due to the presence of reduced thiols (GSH, GSSG) (Santana et al. 2017). Thus, the disturbed redox state, post-sinigrin treatment must have upregulated lipase. Triacylglycerol lipase converts triacylglycerol to fatty acids. However, the metabolomics showed no significant difference in fatty acid levels post-sinigrin treatment. Further, lipases have Ser-His-Asp catalytic triad and serine protease inhibitor PMSF inhibited lipases from R. palmarum (Santana et al. 2017). We observed the upregulation (11 log2 folds) of dipetalogastin-like serine protease inhibitor in the transcriptome. Apart from this, we also observed significant downregulation (7 log2 folds) of the luciferin-4-monooxygenase-like enzyme (Supplementary Data 3). These are the light-emitting enzymes of the adenylate-forming superfamily in fireflies and click beetles (Fraga et al. 2004). Orthologs of the firefly luciferase have a dual function, I) ATP-dependent monooxygenase in bioluminescence II) lipid metabolism, due to its structural similarity to fatty acyl-CoA synthetase. Therefore, the exact role of LPLAT, lipase, serine protease inhibitor and luciferin-4-monooxygenase remain to be explored in sinigrin-treated insects.

Secrets of developmental retardation revealed in the transcriptome

ADAMTs (a disintegrin and metalloproteinase with thrombospondin motif) belong to the family of ADAM proteins. ADAMTs function in the extracellular matrix (ECM) and have diverse roles in invertebrates such as adhesion, proteolytic processes, tissue morphogenesis, and homeostasis (Kelwick et al. 2015). In Drosophila, ADAMTs are involved in cell migration, organogenesis, signalling and development (Tsogtbaatar et al. 2019). In all, ADAMTs are indispensable for tissue remodelling; that is the prime action during metamorphosis. Here, we have observed around 7–eightfold downregulation of ADAMTs, leading to a clue for a reason behind developmental deformities and fatal consequences of sinigrin exposure to H. armigera larvae (Agnihotri et al. 2018).

As a whole, our transcriptomic and metabolomic data suggest that sinigrin-induced oxidative stress in H. armigera is tackled by elevated levels of glutathione and its precursor metabolites of amino acids and fatty acids.

Conclusion and future prospects

This study provides the first comprehensive view of gene expression and metabolic change in insects upon glucosinolate feeding. H. armigera thrives on multiple hosts, having specific dominant secondary metabolites; thus, their system is dynamic, enabling its adaptation to detoxification of undesired metabolites. A study that can target the adaptation of insects precisely towards some metabolites is essential to understand it more thoroughly at the molecular level. Our transcriptomics and metabolomics study of H. armigera describe detailed changes in gene expression and metabolite abundance between control and sinigrin-treated insects. Sinigrin-derived AITC targets mitochondria, leading to respiratory uncoupling that causes harmful effects in insects. The prolonged exposure of glucosinolates causes overall growth retardation in insects. Therefore, the insects must detoxify sinigrin to survive. Consequently, this allows us to propose that the molecular and metabolic flux is inclined towards GSH biosynthesis for glucosinolate detoxification.

It will be interesting to understand the dynamics and functions of H. armigera myrosinases. As myrosinases are plant enzymes, their presence in the current transcriptomic studies raises questions about their evolution and role in insects. The association of myrosinases with different types of glucosinolates and effect of indolic glucosinolates on H. armigera needs to be explored. Myrosinase knockdown followed by glucosinolate feeding will help in pinpointing their role in insect biology. Further, it will be interesting to study the impact of AITC on insect mitochondria and other cell organelles. Also, a detailed proteomics study of ITC-treated insects will further support the idea of mitochondrial dysregulation. Apart from this, it will be intriguing to see the effects of AITC on other cell organelles of insects as it will give insights into their broad-range effects on insect physiology.

Supplementary Information

Below is the link to the electronic supplementary material.

Compliance with ethical standards

Conflict of interest

The authors declare no competing financial interest.